docs: add research notes navigation map

[safal207]

Context

CML now has several useful research notes:

causal invalidity patterns
-> agentic workflow causal boundaries
-> high-value integration opportunities
-> technical report outline
-> benchmark expansion plan

These are valuable, but a new reader may not know the best order to read them.

This issue is a lightweight documentation task: create a research navigation map that helps contributors, grant reviewers, and new readers understand the research path.

Suggested contributor

This could be a good optional follow-up for @Abhishekmystic-KS if you are interested, because your audit findings glossary contribution helped make CML easier to understand in plain language.

No pressure — this is a small docs/onboarding task.

Goal

Add a concise index file:

docs/research/README.md

The file should explain which research note to read first, second, third, and why.

Suggested structure

# CML Research Notes

This directory collects research-facing notes for Causal Memory Layer.

## Recommended reading order

1. `CAUSAL_INVALIDITY_PATTERNS.md`
   - Start here to understand the current CML audit rule families R1-R4.

2. `AGENTIC_WORKFLOW_CAUSAL_BOUNDARIES.md`
   - Read next to understand future agentic workflow boundary patterns.

3. `HIGH_VALUE_INTEGRATION_OPPORTUNITIES.md`
   - Read this to understand where CML-style causal audit might be useful.

4. `TECHNICAL_REPORT_OUTLINE.md`
   - Read this for the grant/research report structure.

## Related evidence docs

- `docs/evidence/BENCHMARK_EVIDENCE_SNAPSHOT.md`
- `docs/evidence/BENCHMARK_EXPANSION_PLAN_50K_100K.md`
- `docs/evidence/EXTERNAL_VALIDATION_PROTOCOL.md`

Acceptance criteria

• docs/research/README.md exists.
• It lists the main research notes in a recommended reading order.
• Each item has a one-sentence explanation of why it matters.
• It links to benchmark/evidence docs.
• It is plain-language and useful for new contributors.
• It avoids broad safety, compliance, or production-readiness claims.
• No runtime code changes are needed.

Non-goals

• Do not rewrite the existing research notes.
• Do not add new claims.
• Do not change CML audit semantics.
• Do not implement new benchmark fixtures.

Why this matters

This improves the contributor and reviewer path:

new reader
-> research index
-> current rule taxonomy
-> future boundary taxonomy
-> integration hypotheses
-> benchmark / validation evidence

That makes CML easier to understand, review, and contribute to.

[Abhishekmystic-KS]

Thanks, I’d be happy to take this. I’ll add docs/research/README.md with a plain-language reading order and links to the related evidence docs, keeping it docs-only and avoiding any new claims or semantic changes.