Correctly infer timezone from event begin time

[stefanv]

Timezones can come from the YAML file (timezone: ...), or from the
event start time: 2025-07-15 17:00:00 +00:00

However it is specified, this timezone should be propagated to the RRULE.

This PR is also a bit more strict, failing when the timezone specifier is unrecognized.

[stefanv]

Needs tests still, but make timezoned events show up in our online calendar.

Not sure whether mixed events (events specified in different timezones) will appear correctly.

[stefanv]

@bsipocz This should be good to go now. Tested with our online calendar, and all looks good.

[bsipocz]

Thank you Stefan!

I also push this back to the test PR in scientific-python/scientific-python.org#782 as I want to double check the generated outputs, not just the widget (as I recall we had issues with both).

[bsipocz]

Unfortunately the defaulting is somewhat buggy, see these two screenshots taken by two different timezone settings on the laptop. Expectation is that all the meetings should be shifted, yet a few are stationary:

[stefanv]

With this PR, the expectation is:

• With no timezone specified, events are "floating" (i.e., local to user's browser timezone)
• Timezone can be set in top-level YAML, or in event, or by using a begin timestamp with +00:00 or similar. This is also reverse order of precedence.

We could make UTC default, but that would be another change.

[stefanv]

I'm concerned about the SPEC Planning Meeting not shifting. It is specified in UTC (+00:00) format. It may be worth Ctrl-Shift-Reloading, I've noticed that calendar changes don't always take without it.

I'm pretty sure I've tested this case, but let's confirm.

[stefanv]

Here's the ICS generated for the SP events; looks correct to me (using this PR), selecting UTC:

BEGIN:VCALENDAR
VERSION:2.0
PRODID:ics.py 0.8.0-dev0 - http://git.io/lLljaA
BEGIN:VEVENT
DTSTART:20250715T170000Z
DURATION:PT1H
SUMMARY:Scientific Python Learn & Lectures sites
UID:[email protected]
DESCRIPTION:Discussions around how to organize learning material 
 for\nScientific Python.\n\nMeeting Link: 
 https://berkeley.zoom.us/my/scientificpython\nMeeting notes: 
 https://hackmd.io/zVZtDmAvRduUY5zg0giMkg?both
DTSTAMP:20251202T192943Z
RRULE:FREQ=MONTHLY;BYDAY=3TU;INTERVAL=1
END:VEVENT
BEGIN:VEVENT
DTSTART:20250902T170000Z
DURATION:PT1H
SUMMARY:Scientific Python Tools
UID:[email protected]
DESCRIPTION:Bimonthly community meeting to discuss and work on tools hosted
  and\nmaintained within the Scientific Python Project.\n\nMeeting Link: 
 https://berkeley.zoom.us/my/scientificpython\nMeeting notes: 
 https://hackmd.io/@bsipocz/SP_tools_meeting
DTSTAMP:20251202T192943Z
RRULE:FREQ=MONTHLY;BYDAY=1TU;INTERVAL=2
END:VEVENT
NAME:Scientific Python Community Calls
X-WR-CALNAME:Scientific Python Community Calls
END:VCALENDAR

[bsipocz]

For the floating SPEC meeting I used the format without the timezone info; which was fixed yesterday. So it's a no concern here.

And we also cleared it up outside of github that the floating meetings is an expected behaviour; I consider it's a behavioural bug (I can't imagine a meeting for which we should allow it); but fixing that; I agree; falls beyond the scope for this PR.

[bsipocz]

Changes here look good. Thanks!