Log exception with crash_reason

[iloveitaly]

Thanks for the amazing work on this package! Super helpful.

I ran into two issues:

• In production, when a job was failing, I wasn't getting any errors logged
• Additionally, none of these failures were being reported to my error monitoring (sentry)

Here's what I found:

• The reason I was't getting any errors logged was because I have my log level set to info. I'd expect most deployments to use a non-info log level.
• An exception from a scheduled job shouldn't be a debug log but instead should use something like error.
• If you are using an exception logging system such as sentry, you need crash_reason to exist in your log metadata for the log message to be treated as an exception. Here's an example implementation in broadway

This change fixes both of these problems.

I've submitted a proposal to elixir-lang to get this function included in the stdlib: https://groups.google.com/g/elixir-lang-core/c/pWz-uTVMEVM

[coveralls]

Pull Request Test Coverage Report for Build 450077e6f9a3a2d438333fa1219f5354a4553312-PR-464

• 4 of 4 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+1.3%) to 86.849%

---

Totals
Change from base Build 3ac7984f4bfca2b47fb9d36e70e56fcaa302dbd6:	1.3%
Covered Lines:	317
Relevant Lines:	365

---

💛 - Coveralls

[maennchen]

Thanks @iloveitaly for the PR!

Could you also add a test for that new log message?

[iloveitaly]

@maennchen done!