Map long and Long arguments to BIGINT

[joschi]

The SQL type INTEGER often (always?) maps to integer types which are only 32 bits (4 bytes) long while BIGINT maps to integer types which are 64 bits (8 bytes) long.

Considering this, mapping the Java types long and Long to the SQL type BIGINT instead of INTEGER seems to make more sense.

This is also consistent with the handling of OptionalLong in Jdbi:

jdbi/core/src/main/java/org/jdbi/v3/core/argument/OptionalArgumentFactory.java

Lines 38 to 44 in ee22bb8

	register(OptionalLong.class, Types.BIGINT, (p, i, v) -> {
	if (v.isPresent()) {
	p.setLong(i, v.getAsLong());
	} else {
	p.setNull(i, Types.BIGINT);
	}
	});

FWIW, this is motivated by a bug we discovered in CockroachDB when sending NULL values in a prepared statement which were sent as NULL::INT4 by Jdbi 3.20.1 instead of NULL::INT8 for a column declared as INT8:
cockroachdb/cockroach#67605

[stevenschlansker]

Thanks for catching this!