tthd analysis rev 2: reattribute trigger to p_dIntlk interlock

Erik Watkins pointed out that BL4B:Mot:shutter:Position transitioned
during the 2026-04-10 23:56 incident. Cross-checking the archiver for
BL4B:Mot:Input:p_dIntlk (ICP DI0, hardware amp-enable gate for axis F
per icp_signals.template:43-50) showed the interlock was asserted from
23:08:19 through 23:55:53.616 — covering every "Encoder stall stop
motor F" event in both retry bursts — and cleared exactly one second
before the 4th retry's BGF succeeded with the accumulated runaway
target. All three prior "stall" clusters (2026-03-27, 2026-03-30,
2026-04-10) are coincident with p_dIntlk=1 windows, so they are not
hardware stalls but interlock-inhibited motion attempts.

Revisions:
- Executive summary: trigger reattributed from encoder stall to
  hardware interlock; keep NewPD1 as root cause of the runaway
- Timeline annotated with p_dIntlk transitions; key-parameters table
  adds p_dIntlk and shutter:Position rows
- New Priority 2: wire p_dIntlk -> p_d.DISP so the motor record itself
  refuses a move when the interlock is asserted (no BGF, no retry)
- Old Priority 3 (hardware investigation of chronic encoder stall)
  retired; there is no p_d mechanical fault to investigate
- Dead ends: add lesson about misreading Galil "encoder stall" on an
  amp-enable-gated axis; it means "amp didn't drive", not a stall
- Files referenced: add icp_signals.template and pps_shutter_pos.db

PDF rebuilt.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>