workspace(perf,refactor) Bulk set-option helper + 10ms readiness polling

[tony]

Summary

Three small, self-contained workspace-builder changes salvaged from a stalled experimental branch:

1. _wait_for_pane_ready polling tightened from 50ms to 10ms. Cuts ~2-3s off suite wall time (within run-to-run noise but consistent with commit body's claim).
2. _bulk_set_options helper added — a single, batch-shaped entry point for the four set-option loops the builder runs (session options, global_options, per-window options, options_after). Today the helper body is a plain loop, so this is a pure refactor; the call sites land on it so the body can later swap to a pipelined dispatch (e.g. a hypothetical Server.batch()) without touching the call sites.
3. Call sites routed onto the helper for the four set-option loops.

The helper exists today as forward-compat plumbing — there's no measurable perf gain from the refactor against current libtmux (0.56.0 ships only the subprocess path; batching pays off on engines libtmux hasn't released). The polling-tighten commit is the only commit that materially affects wall time.

Why this PR exists

The bulk-options helper originated on a local-only experiment branch (libtmux-protocol) that pinned libtmux to a private worktree exposing a Server.batch() API. That experiment hasn't shipped upstream and may take time to land. Rather than let the call-site shape rot on a branch nobody can review, this PR salvages the shape against current libtmux. When Server.batch() (or equivalent) ships, the helper body becomes a one-line internal swap; the call sites stay as-is.

The polling-tighten commit was on the same experiment branch and is independent — it lands here because it shares context.

Measured timing (3 runs salvage, 2 runs master)

Branch    Run  pytest   wall    notes
master    1    76.50s   71.73s
master    2    78.12s   70.79s
salvage   1    67.27s   62.92s
salvage   2    82.19s   74.66s  2 reruns (pre-existing tmux flake)
salvage   3    77.06s   69.41s

Median pytest is essentially tied (~77s). Median wall favors salvage by ~2s, matching the commit body's "~2.5s wall time" claim. One salvage run hit a tmux-state flake with 2 reruns — same flake potential exists on master, just didn't surface in these particular master runs.

Per-commit breakdown

5cef952a — perf(workspace[wait]) tighten _wait_for_pane_ready polling interval to 10ms

One-line change to the interval default in _wait_for_pane_ready. Tighter loop catches the early-readiness window (50-150ms tmux shell-ready latency) more reliably; the previous 50ms interval was missing the leading edge by definition.

6ccb0590 — workspace(feat[builder]) add _bulk_set_options helper for set-option loops

Adds WorkspaceBuilder._bulk_set_options(items, *, target, scope_flag). Mirrors OptionsMixin.set_option's True/False -> "on"/"off" normalisation so loop and helper paths produce identical tmux commands. Reuses libtmux's handle_option_error to preserve the OptionError subclasses callers already catch.

Body is a plain loop today. The API shape (mapping + scope flag + optional target) is deliberately batch-shaped so the body can swap to a pipelined dispatch when libtmux exposes one.

Adds two regression tests (test_bulk_set_options_propagates_unknown_option_error, test_bulk_set_options_applies_session_window_and_options_after).

9d4e4f3f — workspace(refactor[builder]) route set-option loops through _bulk_set_options

Converts the four hot loops to call the helper. Pure refactor — same N round-trips per call site, same observable behaviour. The original commit framed this as a perf change; rewritten here as a refactor since Server.batch() isn't on libtmux 0.56.0 and the original commit body itself notes batching is "no-op cost on subprocess and imsg" anyway. Perf framing deferred to whenever libtmux ships a batching API.

Test plan

• uv run ruff check . --fix --show-fixes — clean
• uv run ruff format . — clean
• uv run mypy — clean (124 source files)
• uv run py.test --reruns 0 -vvv — 799 passed, 2 skipped (master had 797 + 2 new regression tests)
• just build-docs — succeeds
• Per-commit timing comparison vs master — see above

[codecov]

Codecov Report

❌ Patch coverage is 80.95238% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.92%. Comparing base (6edda37) to head (c13b428).

Files with missing lines	Patch %	Lines
src/tmuxp/workspace/builder.py	80.95%	2 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1040      +/-   ##
==========================================
- Coverage   81.98%   81.92%   -0.07%     
==========================================
  Files          28       28              
  Lines        2548     2561      +13     
  Branches      485      487       +2     
==========================================
+ Hits         2089     2098       +9     
- Misses        328      330       +2     
- Partials      131      133       +2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.