Collect/display observing-time warnings/errors

We have long talked about adding a mechanism for observers to quickly see problems while observing, and not just out-of-spec measurements like temperatures or pressures, which are the domain of the STS. A few recent examples:

• drpActor ingest and reduction failures/warnings
• long gen2 frame requests
• opdb lock errors
• r2 amp failure
• various keepalive failures

We propose a minimal and non-disruptive system, where the pfsGUIActor tracks keywords and grows a new panel to display active/recent warnings. It would use the existing yaml configuration and alarm logic as the alertsActor (mainly regexps, but also keepalives), but would not use the alertsActor itself nor its associated keyword annotations. If real alarm notifications are needed (trickled email/text, or fancy latching and clearing), those should still go through STS and the existing alerts.

One goal was to make it easy to add and see warnings without needing to rebuild layers of products, hence skipping ics_actorkeys and ics_alertsActor. We would still use pfs_instdata, but might want to implement that version updating ticket.

Most warnings would use the existing keyword regexp pattern of "OK" vs. some descriptive text intended for humans.

The pfsGUIActor will need to add some latching and acknowledgment logic.