Feat: Update partitioning by DATE, DATETIME, TIMESTAMP, _PARTITIONDATE

[chalmerlowe]

This adds some additional functionality to properly handle partitioning of columns with the following datatypes.

• DATE
• TIMESTAMP
• DATETIME
• _PARTITIONDATE

Where appropriate, ensures the following functions can be used with/or without the following TimePartitioningTypes (HOUR, DAY, MONTH, YEAR).

• DATE_TRUNC()
• TIMESTAMP_TRUNC()
• DATETIME_TRUNC()
• DATE()

This is a nearly complete fix for #1072. The table from #1072 is included here:
NOTE: This PR does not handle _PARTITIONTIME

Column Data Type	HOUR	DAY	MONTH	YEAR
DATE	N/A	Fixed by #1057	via DATE_TRUNC	via DATE_TRUNC
DATETIME	Incorrectly implemented via DATE_TRUNC TODO: use  DATETIME_TRUNC	Incorrectly implemented via DATE_TRUNC TODO: use  DATETIME_TRUNC	Incorrectly implemented via DATE_TRUNC TODO: use  DATETIME_TRUNC	Incorrectly implemented via DATE_TRUNC TODO: use  DATETIME_TRUNC
TIMESTAMP	via TIMESTAMP_TRUNC	via TIMESTAMP_TRUNC	via TIMESTAMP_TRUNC	via TIMESTAMP_TRUNC
_PARTITIONDATE	N/A	N/A	N/A	N/A
_PARTITIONTIME	Not currently implemented TODO: USE TIMESTAMP_TRUNC	Not currently implemented TODO: USE TIMESTAMP_TRUNC	Not currently implemented TODO: USE TIMESTAMP_TRUNC	Not currently implemented TODO: USE TIMESTAMP_TRUNC

[conventional-commit-lint-gcf]

🤖 I detect that the PR title and the commit message differ and there's only one commit. To use the PR title for the commit history, you can use Github's automerge feature with squashing, or use automerge label. Good luck human!

-- conventional-commit-lint bot
https://conventionalcommits.org/

[suzmue]

field == "_PARTITIONDATE" is always true

[suzmue]

I'm a little confused, the type isn't passed to the trunc_fn, should it be?

[chalmerlowe]

Been looking at your comments related to partitioning.

My original code doesn't seem to be 100%, so glad you asked some questions. = )
I am gonna revisit the code and will try to make it correct and simpler, if possible.

Our goal is to increase our coverage of use cases to four.

Within each of those four, the SQL signature will look different depending on the use case and the associated function AND the allowable TimePartitioningType.

Some things to note:

• _PARTITIONDATE does not allow a function OR a TimePartitioningType at all.
• _PARTITIONDATE is a pseudocolumn and is in every table (but not normally visible)
• Some functions only take a couple of TimePartitioningTypes (TPT). See the breakdown below.

# _PARTITIONDATE has no function, no TPT

CREATE TABLE `experimental.some_table` ( `id` INT64, `createdAt` DATE ) # has pseudocol: _PARTITIONDATE
PARTITION BY _PARTITIONDATE;

# DATETIME has function and four TPTs

CREATE TABLE `experimental.some_table` ( `id` INT64, `createdAt` DATETIME )
DATETIME_TRUNC(<datetime_column>, DAY/HOUR/MONTH/YEAR); 
 
# TIMESTAMP has function and four TPTs

CREATE TABLE `experimental.some_table` ( `id` INT64, `createdAt` TIMESTAMP )
TIMESTAMP_TRUNC(<timestamp_column>, DAY/HOUR/MONTH/YEAR);

# DATE has function and only two TPTs

CREATE TABLE `experimental.some_table` ( `id` INT64, `createdAt` DATE )
PARTITION BY DATE_TRUNC(createdAt, MONTH/YEAR);