RUST-807 Disallow maxPoolSize=0

[abr-egn]

RUST-807

This disallows maxPoolSize=0 and merges the relevant updated spec tests for minPoolSize/maxPoolSize, skipping the one for maxPoolSize=0.

[patrickfreed]

So in SWIFT-1339, we decided to not implement this ticket, since we didn't already have this behavior and didn't exactly want to encourage it. What do you all think about doing that here too? cc @kmahar

[abr-egn]

So in SWIFT-1339, we decided to not implement this ticket, since we didn't already have this behavior and didn't exactly want to encourage it. What do you all think about doing that here too? cc @kmahar

The notion of an unbounded pool did make me a bit twitchy, so I'd be happier with this not being implemented.

[isabelatkinson]

Not implementing this is alright with me. Do we need this behavior for tests?

[abr-egn]

Not implementing this is alright with me. Do we need this behavior for tests?

The tests just validate that maxPoolSize=0 is accepted, not what behavior it causes. Before this PR, as far as I can tell the driver would have accepted it and tried to execute with that as the somewhat nonsensical maximum size. I think least surprising behavior from the user's perspective would be to reject it as invalid, which would require us to skip those particular tests.

[kmahar]

I think it's reasonable to take the same route we did in Swift and not treat zero as unlimited.

We did discuss at some point recently-ish what Rust should do on invalid/unsupported URI options and there were some points made in both directions for erroring vs. just logging a warning. I don't remember if we reached a conclusion... Regardless, it would seem we might want to apply similar logic to how we handle invalid values for valid options, such as 0 in this case.

[abr-egn]

Updated the PR and description to disallow maxPoolSize=0; please take another look.