docs(building): a green fast lane can mean nothing ran

The should-run gate reports CI-Test-group-Fast as pass when it skips, so a green check
has two meanings. Record where to look to tell them apart, now that the gate writes its
decision and its comparison base to the job summary.

This commit is deliberately docs-only and deliberately last: with the bug in #2034 this
branch would have skipped build, every Fast shard and Quarkus, and reported green. With
the fix in the previous commit the lane runs, because the branch as a whole touches
ci.yml. The PR is its own regression test.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01D71LBYmQNf5J94wPqr81Jx

Author: jcschaff

docs/BUILDING.md

@@ -157,6 +157,26 @@ assuming it went missing.
 
 See issue [#2038](https://github.com/virtualcell/vcell/issues/2038).
 
+## A green fast lane can mean "nothing ran"
+
+The `should-run` gate in `ci.yml` skips the fast lane when a change touched only
+documentation, and reports the aggregate `CI-Test-group-Fast` as **pass** when it does.
+That is intended, but it means a green check has two meanings: "the tests passed" and
+"there were no tests to run".
+
+To tell them apart, open the `should-run` job. It writes its decision to the run summary:
+
+```
+### CI lane: SKIPPED - nothing was compiled or tested
+Reason: only documentation changed (docs/MESSAGING.md)
+Base for the comparison: `a1b2c3d`
+```
+
+It used to be worth checking this by hand, because the gate judged a whole new branch by
+its final commit — a branch ending in a docs commit skipped everything and looked green
+(issue #2034, fixed). The summary line exists so that class of thing is visible rather
+than inferred.
+
 ## Each git worktree is its own build environment
 
 `git worktree` gives you an isolated checkout, and the build state is isolated with