Add MappedLocalTime::try_map

[pitdicker]

Passing on one error in unix::Cache::offset requires a match that is very similar to the one in TimeZone::from_local_datetime.
I added MappedLocalTime::try_map that returns MappedLocalTime::None that can be used in both places.

For the 0.5.x branch it will be easy to convert this method to return a Result instead.

[codecov]

Codecov Report

Attention: Patch coverage is 75.00000% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 91.82%. Comparing base (3adfd88) to head (e8db5cd).

Files	Patch %	Lines
src/offset/mod.rs	73.33%	4 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1533   +/-   ##
=======================================
  Coverage   91.82%   91.82%           
=======================================
  Files          40       40           
  Lines       18345    18345           
=======================================
  Hits        16846    16846           
  Misses       1499     1499           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[djc]

This PR on its own doesn't look well-motivated... I'd probably keep this commit in the Cache PR but am okay if you want to get it merged first.

[djc]

I think the Option variant of this is called and_then(), maybe reuse that name?

[pitdicker]

👍 and_then() is better.

[pitdicker]

This PR on its own doesn't look well-motivated...

Not sure what you mean. Maybe as another motivation: to implement addition and substraction for a CalendarDuration (#1282) months and days should be added in localtime with a MappedLocalTime as result. Then seconds should be added using a method such as this so the type remains a MappedLocalTime.

But I have to admit I'm not sure if the current two uses in this PR for the 0.4 versions will remain useful for 0.5 with one of the plans in 1448#issuecomment-1987220921. So the method is pub(crate) for now.

[djc]

Ah, sorry, I missed that there are sort of two users for the new method. Looks a little surprising because one of the callers did not previously have the extensive logic implemented in the method.

[pitdicker]

Yes, that was the issue I set out to fix 😄. Sorry for not saying so better.