fix(parsers.avro): Handle timestamp format checking correctly

[athornton]

• Updated associated README.md.
• Wrote appropriate unit tests.
• Pull request title or commits are in conventional commit format

resolves #13854

Made check for invalid timestamp format happen outside the if clause that only fired if the format was the empty string.

[athornton]

We might want to dump this one in favor of #13856, which subsumes it (handling the format correctly turned out to be useful for testing that fix).

[srebhan]

Thanks for the fix @athornton! Just one small comment...
I think we should keep this fix in a separate PR as the other PR also changes the behavior of the metric output and that might be controversial...

[athornton]

Updated, test cases changed, review re-requested.

[srebhan]

Looks good to me. Thanks @athornton!

[powersj]

I am hesitating hitting merge due to the 386 test failure. Do we really need to increase the timeout or is this branch behind master? Or something else?

[athornton]

We're not behind master anymore, and the error seems to indicate we really do need longer:

INFO [config_reader] Config search paths: [./ /go/src/github.com/influxdata/telegraf /go/src/github.com/influxdata /go/src/github.com /go/src /go / /root] 
INFO [config_reader] Used config file .golangci.yml 
INFO [lintersdb] Active 33 linters: [asasalint asciicheck bidichk bodyclose depguard dogsled errcheck errname errorlint exportloopref gocheckcompilerdirectives gocritic goprintffuncname gosec gosimple govet ineffassign interfacebloat lll makezero nakedret nilerr nolintlint prealloc predeclared revive sqlclosecheck staticcheck tenv tparallel unconvert unparam unused] 

Too long with no output (exceeded 10m0s): context deadline exceeded

[powersj]

I find that hard to believe given all you did was modify a switch statement ;)

Looking at master you can see we did not max out the system at all:
https://app.circleci.com/pipelines/github/influxdata/telegraf/17914/workflows/307b4206-1dfd-44cd-94d2-3eaa5e3868a1/jobs/276828/resources

I'm going to land this and hope this is just something strange with the lsst-sqre org.

[athornton]

It may well be something strange with lsst-sqre, given that we moved away from CircleCI and probably dropped from a paid tier to their free one (we found GitHub Actions to be a much more pleasant experience).