Do not write each SQL error multiple times in TestScript

Improve test stability

Fix TestDateTimeUtils.testDayOfWeek() and example with ANY(?)

Reimplement DATEDIFF without a Calendar

Add support for = ANY(?)

Allow to read invalid February 29 dates with LocalDate as March 1

Encode date and time in fast and proper way in PgServerThread

Reimplement remaining parts of EXTRACT, ISO_YEAR, etc without a Calendar

to see if we can figure out why it fails on travis

because despite setting it in the global settings, maven still doesn't pick it up

Read known fields directly in DateTimeUtils.getDatePart() V3

Inline getTimeTry() into DateTimeUtils.getMillis()

Add simple implementations of isWrapperFor() and unwrap() to JdbcDataSource