Add greppable markers to CoreLog for all microscope geometries; elapsed time report even during errors

[bls337]

This PR restores functionality to LSM from the 1.4 diSPIM plugin:

• LSM measures the same timing window as the original plugin
• LSM also reports the elapsed time on errors which if failed to do before

There are some known errors that relate to hangs in run(), the solution for those problems is more involved and will be revisited when working on other issues, so a hang in run() wont show LSM-ACQ-STOP right now.

The LSM-ACQ-START / LSM-ACQ-STOP markers bracket every acquisition, so one sed pulls a single run's window straight out of a CoreLog, no tooling required:

sed -n '/LSM-ACQ-START 4 /,/LSM-ACQ-STOP 4 /p' CoreLog.txt

If you want line numbers you can use grep.

---

Because LSM-ACQ-STOP times the whole acquisition window (not just the imaging body), per-mode overhead is obvious at a glance. From one five-run session:

LSM-ACQ-STOP 1 3636 ms   # Galvo scan
LSM-ACQ-STOP 2 2896 ms   # Galvo scan
LSM-ACQ-STOP 3 2890 ms   # Galvo scan
LSM-ACQ-STOP 4 8505 ms   # Stage scan  <------------- ~5.6 s longer!
LSM-ACQ-STOP 5 2847 ms   # No scan (fixed sheet)

The stage-scan run is ~5.6 s longer than the others, an extra delay in the stage-scan cleanup path that's easy to miss but jumps right out once each run's full window is timed. (The old per-engine "acquisition took" line measured only the imaging body, so it hid this.)

Surfaced a stage-scan cleanup delay; fix in the next PR, this was discovered as a result of the #404 CoreLog that @oleksiievetsno posted.