'Alert' helper and accessible log

It's all too easy for messages in the EDMC status text to be overwritten by something else before you get chance to read them, let alone react to them.

- [ ] Make a helper for setting them, rather than the **very** ad hoc "just change the text directly" that's scattered all over.
- [ ] Enhance this with the ability to specify, essentially, a log level. Match the available ones to those in `logging` and then we could also have this *also* literally log the message.
- [ ] Now, for anything that's at an appropriate level (probably WARN or higher, but maybe INFO and up), add it to record. We *might* want to actually just have this be another logged file. Whether it could literally be another `logging` channel should be considered, even if the content then needs some processing for the next point. Having this be a literal file would naturally offer persistence across restarts if desired (or we can truncate it at the start of each invocation).
- [ ] Add a mechanism to acknowledge any 'alert'. If we log anything below WARN level then this should probably only be WARN and up. If there are any unacknowledged alerts then display a `!` or similar icon ... far left of the status line ? Clicking this should open a UI for reviewing, marking read, etc, the extant alerts.

Start by locating the scattered EDMC status-text updates and the existing Python logging usage. The issue spans alert levels, persistence, acknowledgement, and a review UI; done would require an agreed design covering those behaviors and confirming that unacknowledged alerts are visible and reviewable.

Written by the indexing model from the issue text.