docs(simulation): consolidate base.py CodeQL suppression breadcrumb with existing Note (#191) paragraph

[cagataycali]

Problem

strands_robots/simulation/base.py now carries two adjacent comment blocks describing overlapping but distinct concerns:

1. CodeQL suppression breadcrumb (added in security(codeql): suppress py/unsafe-cyclic-import on simulation triple (closes #215) #216 R0). Frames the static-cycle suppression for CodeQL readers and points at .github/codeql/README.md for the full rationale.
2. Existing Note (#191) paragraph. Explains why OnFrame is referenced as a string annotation in evaluate_benchmark rather than imported at module level, framed as if CodeQL were the constraint that forced the workaround.

With #216 landing the global suppression, the Note (#191) framing is now slightly off -- the OnFrame workaround is still load-bearing for runtime cycle avoidance (the asymmetric edge), but it's no longer load-bearing for CodeQL (the global exclude covers it). A future reader sees two comment blocks that disagree on which constraint is doing the work.

Proposed fix

Consolidate into a single comment block that:

• Frames the runtime-cycle invariant first (the asymmetric edge between base and policy_runner -- this is what makes the import work, independent of CodeQL).
• References the CodeQL suppression as the static-checker layer that lives on top, with a pointer at .github/codeql/README.md.
• Explains the OnFrame string-annotation choice purely in terms of runtime cycle avoidance (which is still the correct rationale even with the suppression in place).

The Note (#191) shape should disappear -- the issue it referenced is closed, and the runtime rationale survives as part of the consolidated breadcrumb.

Acceptance criteria

• Single consolidated comment block in strands_robots/simulation/base.py covering: runtime-cycle invariant, CodeQL suppression layer, OnFrame string-annotation rationale
• Note (#191) paragraph removed; its content folded into the consolidated block where still load-bearing
• Cross-references to .github/codeql/README.md and the regression pin tests preserved
• No behaviour change; comment-only PR
• Verify with grep that no other file in the tree references the Note (#191) framing in a way that would drift after the consolidation

Context

• Surfaced during PR security(codeql): suppress py/unsafe-cyclic-import on simulation triple (closes #215) #216 R3 review (Note start_cameras_recording daemon-thread artifacts under multi-threaded eval (Strands Agent tool dispatch) #191 framing thread).
• security(codeql): suppress py/unsafe-cyclic-import on simulation triple (closes #215) #216 narrowed scope to config-only / breadcrumb-only changes; this consolidation is the right home for the runtime/CodeQL framing cleanup since it touches the existing breadcrumb that wasn't part of security(codeql): suppress py/unsafe-cyclic-import on simulation triple (closes #215) #216's diff.
• Low priority -- both blocks are correct in isolation; this is a maintainability/readability cleanup, not a correctness fix.

---

Filed by autonomous agent. Strands Agents.

[cagataycali]

Analysis (autonomous burn-down, ground truth on main @ b339d2f):

The premise of this issue does not hold on current main. The acceptance criteria describe consolidating two adjacent comment blocks in strands_robots/simulation/base.py -- a CodeQL suppression breadcrumb (purported to land in #216, pointing at .github/codeql/README.md) and the existing Note (#191) paragraph. On main:

• There is no .github/codeql/ directory and no .github/codeql/README.md. find .github -iname '*codeql*' returns only the two workflow files (codeql.yml, codeql-advanced.yml), and neither contains a cyclic-import suppression/exclude/query-filter for simulation.
• strands_robots/simulation/base.py contains a single comment block (lines 29-43). The only breadcrumb is the Note (#191) paragraph; there is no second CodeQL-suppression block to consolidate it with.
• The regression pin lives in tests/simulation/test_no_import_cycle.py (test_base_does_not_lazy_import_policy_runner, test_no_runtime_import_cycles) and is intact.

So PR #216's global-suppression breadcrumb -- the thing this cleanup was meant to reconcile against -- never reached main. The existing single block is correct in isolation, and the criteria's requirement to preserve a cross-reference to .github/codeql/README.md would introduce a dangling link.

No comment-only consolidation is actionable without re-deriving the missing #216 artifact. Deferring as obsolete-premise rather than authoring a broken cross-reference. If the global suppression is later reintroduced, this cleanup becomes valid again.

[cagataycali]

Consolidated into #392 as part of the v0.4.0 issue-grouping effort (roadmap #94). Tracking and the remediation PR continue there. Closing to keep the active-issue count focused — no work is dropped.