Remove assert in production code

[kivel]

The changes will remove assert statements and replace them with check/raise.
Logic should remain identical, except for velocity related assertions in the timeout calculation.

new ruff rule

rule S101 was added to avoid future use of assert

Tango

There are a few remaining uses of assert in the Tango code. I ask for a Tango developer to take a look at those. If fail understand why they are used and also struggle with seemingly unrelated tests failing when I touch them.

assert in general

I understand that a simple assert foo.bar will make language support tools happy, as it is certain that bar exists for subsequent code. There are alternatives, admittedly, not as straight forward, though.
The general question for code hygiene should be, what comes first, a happy type checker or readable code. It happened a few time in my quest to remove the assert statements that it was unclear to me, why assert was used in the first place. I'd at least add a comment, if assert is used to make Pylance happy.

velocity checks

Previously, the velocity needed to be positive. The error message was stating however that the velocity is zero. Since the EPICS motorRecord will work with both, the check was changed to specifically check for zero. Technically, the check isn't even needed, as the subsequent timeout calculation will throw a DivisionByZero exception. This is of course, in the case we allow negative velocities ... which we should.

[kivel]

@coretl I addressed your requests to my abilities.

[coretl]

Please could you resolve the conflicts and remove that commented out code and then I will merge

[kivel]

Please could you resolve the conflicts and remove that commented out code and then I will merge
sure!

merged main into feature and fixed all conflicts 34e0b70

[coretl]

Looks like it still has conflicts? Also, lint seems to be failing...

[kivel]

I have absolutely no idea why the ruff format step in the workflow insists on reformatting the 4 files and make it fail.
It looks to me as if a different formatting rule is used in the workflow. All the tools work on my host in the provide dev-container

(venv) root ➜ /workspaces/ophyd-async (710_remove_assert) $ ruff format
178 files left unchanged
(venv) root ➜ /workspaces/ophyd-async (710_remove_assert) $ pre-commit run -a
check for added large files..............................................Passed
check yaml...............................................................Passed
check for merge conflicts................................................Passed
fix end of files.........................................................Passed
lint with ruff...........................................................Passed
format with ruff.........................................................Passed
Ensure import directionality.............................................Passed
(venv) root ➜ /workspaces/ophyd-async (710_remove_assert) $ git status
On branch 710_remove_assert
Your branch is up to date with 'origin/710_remove_assert'.

nothing to commit, working tree clean

[coretl]

I have absolutely no idea why the ruff format step in the workflow insists on reformatting the 4 files and make it fail.

I think that is because ruff changed its formatting rules (looks like it does that once a year): https://github.com/astral-sh/ruff/releases/tag/0.9.0

If you run pip install ruff --upgrade in your venv then ruff format then it should match CI

[kivel]

@burkeds Thanks for looking into the Tango part.

@coretl Thanks for pointing out the need to update ruff. That made the CI work.
That should conclude this PR.