Handle PG::ConnectionBad from PG#server_version similarly to version 0

[joshuay03]

Motivation / Background

Follow-up to #54713 and ged/ruby-pg#632.

Detail

Now that in future versions pg will raise PG::ConnectionBad when attempting to get the database version with a bad connection, we should handle it the same as when a version of 0 returned.

I opted to not preserve the error message from pg cause it's a bit too specific to the native implementation. Getting the message across to the user that the root of the failure is a bad connection should suffice imo.

cc @byroot

Additional information

N/A

Checklist

Before submitting the PR make sure the following are checked:

• This Pull Request is related to one change. Unrelated changes should be opened in separate PRs.
• Commit message has a detailed description of what changed and why. If this PR fixes a related issue include it in the commit message. Ex: [Fix #issue-number]
• Tests are added or updated if you fix a bug or add a feature.
• CHANGELOG files are updated for the changed libraries if there is a behavior change or additional feature. Minor bug fixes and documentation changes should not be included.

[byroot]

Isn't it already handled by with_raw_connection?

[joshuay03]

Isn't it already handled by with_raw_connection?

Ah yes, however that raises ActiveRecord::ConnectionNotEstablished and will preserve the original message, what do you think about that?

If that's fine, should we update the explicit version 0 raise to match?

[byroot]

If the error message isn't good, we can handle that centrally in translate_exception_class

[joshuay03]

If the error message isn't good, we can handle that centrally in translate_exception_class

I've actioned this and made the version 0 logic raise a PG::ConnectionBad with a message partially matching the underlying one, which I think is fine for this special case? Mainly because the intention behind #translate_exception_class seems to be native exception translation only.

Alternatively we just keep the duplication.

[matthewd]

I really don't think we should be rewriting error messages here, the original message is fine

[joshuay03]

Okay I've updated, all this does now is ensure the error classes are consistent.