use logging instead of sentry.capture_*

[osmaa]

Replace sentry integration's capture_message and capture_exception with Python logging and Sentry's built-in integration to it. This has the following benefits:

• improves logs when Sentry is not used
• improves contextuality, as logging.info() become Sentry breadcrumbs
• reduces footprint of the Sentry-specific code (though scope setup/clear remains)

[andrewgodwin]

Thanks - seems sensible enough!

[kfdm]

Sorry for the drive by comment, but I'm curious about something.

Typically if you use something like logging.info then it logs everything to the root logger. While not specifically an issue, often it seems better to get an instance of a logger so that the logs also show where in the file things are.

import logging
logger = logging.getLogger(__name__)
...
logger.info('log like usual')

[andrewgodwin]

Yes, that's true, but Sentry's integration has a bit more locality due to the breadcrumb behaviour, so while this PR isn't perfect, it's still an improvement over the previous code!