SCHEDULING CLARIFICATION RECORDED. Holding.

  commit: 287fd9d0d68230b264f99ba492e8bcee7ee94707   (branch main, UNPUSHED)
  blob:   2bef2214aaba155af176b41d58d3b0a65366b217
  sha256: 480d54266cad3426c42886b608f15a107e4b324ef586ba3d8847a032989b07d8
  parent: b2b60557. traceable-reqs exit 0, tree clean, section 3 only.

TWO SEPARATE BUDGETS AND A TOTAL, as you ruled:

  observation budget   300 s   all experimental activity, from run start
  cleanup budget       120 s   SEPARATE, begins when observation ends FOR ANY REASON
  total bound          420 s   run start to done

THE DEADLINE ENDS EXPERIMENTAL ACTIVITY AND STARTS CLEANUP -- IT NEVER SKIPS IT. Exhausting the
observation budget is a NORMAL ENDING, not an error, and it never licenses skipping or shortening
cleanup. Worth saying why your catch matters: my previous wording left cleanup's standing ambiguous
at exactly the moment it matters most, since the run that overran is the run most likely to have left
something behind.

EVERY OBSERVATION AND REATTACH PHASE IS ADDITIONALLY CAPPED BY THE REMAINING OBSERVATION BUDGET. A
phase's nominal cap is an UPPER BOUND, NOT AN ENTITLEMENT -- with 40 s left of the 300 s, that phase
gets 40 s. A phase cut short is REPORTED AS TRUNCATED, naming the phase and the time it actually got.
A truncated phase's result is never presented as a completed observation, and its absence of an event
falls under the section 8 "not observed" rule rather than reading as a negative finding.

No other change to the plan. Still prepared, NOT admitted.

Holding. No runtime, no instrumentation, no further rounds from me. W2 run 20260912T231404Z retains
the window through cleanup disposition.