Map inject: catch a quest area's PAC when it is read again #32

A map edit pushed into a quest area has to survive the area's reload: the visible mesh is built from the stage PAC when the area loads, and leaving and re-entering an area evicts the PAC and reads it again. studio map inject --catch (and the research stage_inject before it) arms a write watchpoint on the tail of the PAC's current slot, so it can re-apply the edit while the loader is stopped there.

It has never worked in game:

• Freeze (mitigated in feat(studio): the map workspace and the stage backend #30). In the live test of feat(studio): the map workspace and the stage backend #30, the snow camp's slot was reused after walking camp → section 1. Something wrote the watched word continuously, and the watch stopped the CPU on every write until its deadline. feat(studio): the map workspace and the stage backend #30 now ends the watch on the first foreign write, so the catch reports "not caught" instead.
• Re-read never caught. Nothing shows that a re-read PAC returns to the same slot, so a catch armed on the old address may simply never fire.

To do:

• Find where the engine assigns a slot to a stage PAC it is about to (re)load, and when that happens relative to the read: the resource table mhfu.stage.resident_files reads, or the loader call.
• Arm the catch from there: on the slot the PAC is read into, before the read.
• Prove it live: push a camp edit, walk camp → section 1 → camp, and see the edit drawn after re-entry. The way back is mhfu rig goto (feat: hybrid ports wave 2: studio clips, walking, hit push, every clip in game, engine walls #52): one area change straight to a named point.

Done when a quest-area edit is drawn after leaving and re-entering its area, with no restart.