Project list and workspace details show different conversation counts; workspace load still slow on large installs #95

Problem

After #88 (lazy-load + summary endpoints), two correctness issues and one performance gap remained on large Cursor installs:

1. Count mismatch — The project list page showed a different conversationCount than the number of tabs on the workspace details page for the same project.
2. Noisy logs — Loading workspace summaries logged warnings for composerData:empty-state-draft when the global KV row had a NULL payload (the JSON object must be str, bytes or bytearray, not NoneType).
3. Slow first paint — On machines with many workspaceStorage folders, project list and sidebar summary took ~50s because composer ownership was rebuilt sequentially from local state.vscdb files, and some paths still over-fetched global composerData.

Expected behavior

• Project card conversationCount matches GET /api/workspaces/<id>/tabs?summary=1 tab count for the same project.
• Cursor UI placeholders (e.g. empty-state-draft) are skipped without decode warnings.
• Project list and sidebar summary use workspace-scoped I/O with cached composer registry; first conversation auto-loads after sidebar render (~2s is acceptable).

Root cause (count mismatch)

Project list briefly counted conversations from local allComposers without applying the same global filters as the summary path (non-empty fullConversationHeadersOnly, composer validation, exclusion rules including model names).

A secondary regression dropped conversations during assembly: get_project_from_file_path was accidentally removed while refactoring workspace_resolver.py, causing NameError and parse/assembly warnings on large installs.

Resolution (branch fix/workspace-list-load-and-count)

Shared composer scan

• Extracted services/workspace_composer_scan.py so list and summary paths share the same parse, exclusion, and workspace-assignment logic.

Count alignment

• list_workspace_projects now applies the same filters as list_workspace_tab_summaries (headers, validation, exclusion rules, assignment).

Performance

• Parallelized build_composer_id_to_workspace_id via ThreadPoolExecutor.
• Bulk load_project_layouts_map instead of per-composer N+1 queries.
• Reused mtime-keyed disk cache (summary_cache.py) for project list and tab summaries.

Correctness

• Null/placeholder composerData rows skipped without warnings.
• Restored get_project_from_file_path in workspace_resolver.py.

UI

• Workspace page auto-selects the first tab when no ?tab= query param is present.
• Deep links from search (?tab=<composerId>) fetch the requested conversation directly instead of falling back to the first sidebar item.

Tests

• tests/test_workspace_list_count_alignment.py — count parity
• tests/test_project_path_boundary.py — path helper / assignment regression
• Updated parse-warning and API tests

Acceptance criteria

• conversationCount on project cards equals summary tab count for the same workspace
• No warnings for null/empty composerData placeholders
• Parallel local registry build + scoped global fetches for list/summary paths
• Regression tests for count alignment, workspace scope, and null payload handling
• Workspace UI auto-selects first tab when no ?tab= query param

Out of scope (tracked separately)

Search performance and indexing work landed on the same branch but is not part of this issue:

• 30-day default search window + all_history opt-in
• Local FTS search index (search_index.sqlite) with background refresh
• Search bubble-scan optimization and deep-link workspace assignment fixes

Those should be referenced in the PR and/or a dedicated search issue.

Verify

1. Restart python app.py.
2. Home page: project counts match sidebar totals per project.
3. No “could not be fully assembled” banner from placeholder composers.
4. Open a workspace without ?tab= — first conversation loads automatically.
5. pytest tests/test_workspace_list_count_alignment.py tests/test_api_workspaces.py