Minor JDBC driver improvements

[electrum]

Fixes #6929

[ebd2]

2 minor comments/questions:

1. Will this return 0 for every snapshot version when getDatabaseMinorVersion() is called? If so, that seems like a pain for us or anybody doing testing.
2. For end users, it seems likely that the only reason you'd hit this is because the presto versioning scheme has radically changed and the driver version parsing code is out of date. It seems like a good bet that the user needs to update the driver. How about throwing an SQLException containing the full version string like "Driver version x.y could not parse server version string '%s'. Try updating the driver"

Please feel free to ignore if this is spec-compliant behavior as-is.

[electrum]

1. If the driver JAR is built from Maven, it will work correctly, since the snapshot JAR will have the correct manifest. I just tested this out (and fixed a bug). The version for snapshots is a bit weird -- it's the output of git describe HEAD. So, for this branch on 0.162-SNAPSHOT, the version is 0.161-115-gea22394b5 (and that depends on having fetched the tags to the local Git repo). Thus, you get 0/161 for major/minor, and the full string for getDriverVersion().
2. I considered that, but it will break for stuff like presto.version=testversion that we have in the development config.properties file. I'd rather be conservative for now.

[electrum]

I just realized my comment for #1 is actually about the JDBC driver version, not the server version. However, the answer is nearly the same. The server version is determined the same way, from the package manifest, unless explicitly overridden with presto.version.

[ebd2]

Looks basically reasonable to me % usability nit. Somebody who is more up to date with what is supported should probably look over those changes.

[dain]

Looks good