docs: ROADMAP Done table omits the #178 and #175 completions #202

Evidence

ROADMAP.md's Done table (verified on main 02a5b122b2) has no row for two items completed and closed on 2026-10-20 — grep -n '178\|175' ROADMAP.md returns nothing:

• test(core): mcpxDaemonAutostart fail-open branches (settings exec error / missing binary / timeout) have no test #178 — test(core): mcpxDaemonAutostart fail-open branches have no test — CLOSED via PR test(core): cover mcpx daemon autostart fail-open branches #186 MERGED (b8c91cb82c, test(core): cover mcpx daemon autostart fail-open branches).
• Stale node_modules FOD hash at pin f9aa962 (same class as nix-hashes CI failures) #175 — Stale node_modules FOD hash at pin f9aa962 — CLOSED via PR fix(nix): update stale node_modules hashes and fix the nix-hashes updater #189 MERGED (be4271e50e, fix(nix): update stale node_modules hashes and fix the nix-hashes updater).

The Done table does record same-day completions #183, #185, #180, #144, #170, #155, #142, #174, #167, #166 and #165, so these two are omissions, not a cutoff. The Sources line already claims coverage of 'the merged PRs of 2026-10-20'.

Note #196 was reopened after PR #198 and stays OPEN (its remainder is #199), so it correctly has no Done row — only #178 and #175 are missing.

Why it matters

The ROADMAP is the fork's record of what is done; missing rows make completed work look open and invite duplicate effort (this is the same omission class fixed repeatedly today, most recently #187/PR #188).

Acceptance criteria

• Done table gains one row for test(core): mcpxDaemonAutostart fail-open branches (settings exec error / missing binary / timeout) have no test #178 (mcpx daemon autostart fail-open test coverage, 2026-10-20, result test(core): cover mcpx daemon autostart fail-open branches #186 or test(core): mcpxDaemonAutostart fail-open branches (settings exec error / missing binary / timeout) have no test #178 matching the table's existing convention of linking the PR/issue that completed it) and one row for Stale node_modules FOD hash at pin f9aa962 (same class as nix-hashes CI failures) #175 (stale node_modules FOD hashes updated, 2026-10-20, fix(nix): update stale node_modules hashes and fix the nix-hashes updater #189 or Stale node_modules FOD hash at pin f9aa962 (same class as nix-hashes CI failures) #175).
• No other rows changed; table formatting (prettier) still passes.