Add sentry integration

[MaxAlex]

Adds Sentry integration to Odin. In theory, this just requires adding a sentry_sdk.init() call and everything will just work. There are a few complications:

• Sentry initialization modifies exception handling infrastructure in the current Python process, so this doesn't carry over into subprocesses (such as those used to run OdinJob via multiprocessing.) There's a couple ways to address this, of which the simplest is to just call have OdinJob call init() for every subprocess.
• Sentry ingests errors in two ways; either by parsing error output from logs, or by explicit calls to functions like sentry_sdk.capture_exception. The problem with the former method is that Sentry's error parsing doesn't seem to be very good, and doesn't seem to like our current logging infrastructure, so a single exception will wind up smeared into a dozen or so Sentry incident reports. This would be difficult to fix without significantly changing how logs appear in Splunk, so to keep things consistent I've put in a try-except that calls sentry_sdk.capture_exception in job.py, and disabled log-parsing-based Sentry reports there. Importantly, errors outside of an OdinJob will still be reported from logs.
• cubic_archive.py would previously report missing snapshot data for a table by throwing an exception. Since some tables aren't being sent over in dev this has been constantly emitting alerts, which we've been ignoring. It would also be a constantly-updated Sentry report, so this PR suppresses that 'error.'