Handle full timestamp ranges for redshift, postgres

[dlawin]

Update redshift/postgres normalize_timestamp to handle all possible values
https://www.postgresql.org/docs/current/datatype-datetime.html

• Technically rounds BC timestamps the wrong way
• Edge case at the upper/lower bounds, rounded to keep below overflow
  • the least() and greatest()

Resolves #786

[dlawin]

example query:

SELECT "id"::varchar, CASE WHEN "timestamp_value" IS NULL THEN NULL ELSE  RPAD(LEFT(left(to_char(least('294276-12-31 23:59:59.0000'::timestamp, greatest('4713-01-01 00:00:00.00
                  BC'::timestamp(6), least('294276-12-31 23:59:59.0000'::timestamp(6), "timestamp_value"::timestamp(6))))+ interval '0.0000005', 'YYYY-mm-dd
                  HH24:MI:SS.US'),length(to_char(least('294276-12-31 23:59:59.0000'::timestamp, greatest('4713-01-01 00:00:00.00 BC'::timestamp(6), least('294276-12-31 23:59:59.0000'::timestamp(6),
                  "timestamp_value"::timestamp(6))))+ interval '0.0000005', 'YYYY-mm-dd HH24:MI:SS.US')) - (6-6)), 26), 26, '0') END FROM "dev_dan"."timestamp_test_cases" WHERE ("id" >= 1) AND ("id"
                  < 12)