test: investigate timeouts in go test -race

[charlieegan3]

In #1146, a polling trigger for workspace diagnostics was added in order to stop an issue that occurred only when running the tests in the race detector.

The issue was observed such that a number of language server tests would intermittently run in such a way that events did not leave the cache in the expected state. This lead to timeouts where even after a long delay the results would be incorrect.

This issue was addressed by adding the WorkspaceDiagnosticsPoll option to the creation of new language servers which is only set when running with the race detector on. This is not ideal but eventing functionality is still tested in the regular tests with no poll interval.

This issue is for when there is more time to work out what could be causing this, and address the issue without needing polling in the race detector.