fix: compute Experiment start/end time as union across all devices

[binary69]

Closes #134

Problem
In _load_devices(), start_time and end_time were assigned with plain assignment inside a loop, so they reflected only the last loaded device instead of the global range across all devices.

Fix

changes experiment.py
Updated _load_devices() to use min/max to track the earliest start and latest end across all devices.
Devices with None, NaN, inf, -inf timestamps are skipped with warning, not silently corrupting
ValueError if no device with valid time ranges are loaded
ValueError if compute start_time > end_time

New Tests

create_experiment.py
contextmanager fixture with configurable signals, sampling rate, and override_meta for invalid metadata testing
builds per-device interpolation config dicts

test_experiment.py
parametrized across 1, 2, and 3 device scenarios and signal counts - (5+1)*2 = 12 runs
parametrized across 9 invalid metadata cases - 9 runs
Verifies that start_time > end_time raises ValueError - 1 run
verifies that an invalid device is excluded and a warning is logged, while a co-loaded valid device initialises the experiment correctly. - 9 runs

All 31 tests are passing.

[gitnotebooks]

Review these changes at https://app.gitnotebooks.com/sensorium-competition/experanto/pull/137

[Copilot]

Pull request overview

Fixes Experiment.start_time / end_time so they reflect the global time span across all loaded devices (earliest start + latest end), matching the class docstring and avoiding “last device wins” behavior.

Changes:

• Update Experiment._load_devices() to compute start_time/end_time via min/max across devices.
• Add new tests covering 1-, 2-, and 3-device scenarios for the computed experiment time span.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
experanto/experiment.py	Compute experiment start/end time as min/max across loaded devices.
tests/test_experiment.py	New regression tests for experiment start/end time spanning multiple devices.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[pollytur]

We typically separate the test data creation and the actual testing - please do the same here :)

Ideally look at PR #113 and name the files like create_experiment.py‎ and test_experiment.py such that we could merge your tests together later

[binary69]

Thanks for the feedback! I'll restructure the tests following the pattern in PR #113 — separating the data creation helpers into create_experiment.py and keeping the tests in test_experiment.py. Will push the update shortly.

[Copilot]

Pull request overview

Fixes Experiment.start_time/end_time so they reflect the global time span across all loaded devices (earliest start + latest end), instead of only the last device processed.

Changes:

• Update Experiment._load_devices() to track start_time via min(...) and end_time via max(...) across devices.
• Add new tests covering single-, two-, and three-device scenarios for the computed union time range.
• Add a small test helper to generate multi-device experiment folders under tmp_path.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
experanto/experiment.py	Computes experiment-wide start/end as min/max across device ranges.
tests/test_experiment.py	Adds regression tests validating union start/end time behavior.
tests/create_experiment.py	Adds helpers to build multi-device experiment fixtures for tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Fixes Experiment.start_time / end_time so they reflect the union of time ranges across all loaded devices (earliest start, latest end), aligning behavior with the class docstring and preventing “last device wins” behavior.

Changes:

• Update Experiment._load_devices() to track experiment-wide start_time/end_time via min/max across devices.
• Add warnings/validation when devices have undefined time bounds (though current placement/logic needs adjustment).
• Add unit tests covering single-, two-, and three-device unions using new test fixtures.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
experanto/experiment.py	Compute experiment time range as min/max across devices; adds (currently flawed) validation/warnings for missing/invalid ranges.
tests/test_experiment.py	Adds tests asserting Experiment.start_time/end_time represent the union across devices.
tests/create_experiment.py	Adds helpers to generate minimal on-disk experiment/device fixtures for the new tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Fixes Experiment.start_time / end_time so they reflect the global time span across all loaded devices (earliest start + latest end), rather than being overwritten by the last device processed.

Changes:

• Update Experiment._load_devices() to compute experiment-wide start_time/end_time via min/max across devices (skipping devices with undefined bounds).
• Add a small post-load validation/warning block for missing devices / invalid computed range.
• Add new tests covering single-, two-, and three-device union behavior (with new test fixtures for building a minimal experiment folder).

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
experanto/experiment.py	Compute experiment-wide time range across devices and add validation/warnings.
tests/test_experiment.py	Add regression tests for experiment-wide start/end time union behavior.
tests/create_experiment.py	Add helpers to create minimal on-disk device folders + modality config for tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[binary69]

Addressed the changes abovementioned and expanded test coverage with additional edge cases, including parametrized invalid metadata tests, a dedicated inverted range test, and caplog assertion to verify warnings are emitted when invalid devices are skipped. All 31 tests are passing. Looking forward to your feedback :)

[pollytur]

@binary69 could we please merge main into the current branch?

[binary69]

Thanks @pollytur :) will be updating soon

[binary69]

Made the suggested changes, happy to get more feedback :)

[pollytur]

LGTM, thanks!
@schewskone please take a look as well :)

[schewskone]

LGTM as well. I'm not 100% sure about randomness in parameter generalization for testing but I also don't mind.
Thanks for the contribution

[binary69]

Thank you so much, and I really appreciate the feedback. On the randomness concern, we can either use seeded randomness or fixed non-default valued to avoid potential flakiness. Looking forward to contribute more :)