expression: refine date_add/sub return type and precision (#35009)

[ti-srebot]

cherry-pick #35009 to release-5.3
You can switch your code base to this Pull Request by using git-extras:

# In tidb repo:
git pr https://github.com/pingcap/tidb/pull/35447

After apply modifications, you can push your change to this PR via:

git push git@github.com:ti-srebot/tidb.git pr/35447:release-5.3-4ae78cdead54

---

What problem does this PR solve?

Issue Number: close #31799, close #9813. And part of #31867 will be resolved by this PR.

Problem Summary:

There exists several issues about function date_add and date_sub:

1. Return types of several signatures don't match MySQL's (doc):
   1. date_add/sub(int/double/decimal, interval): MySQL returns string, TiDB returns datetime.
   2. date_add/sub(duration, interval) when interval unit contains YMD part (except for DAY_MICROSECOND - I believe it's a mismatch between MySQL's actual behavior and its document, but it doesn't matter much), MySQL returns datetime with current date padded, TiDB returns time.
   3. Precision of the return type doesn't match MySQL's in many cases. Note that here I mean the precision concluded by type inference - MySQL client option --column-type-info is your good friend.
2. Runtime doesn't respect the precision inferred in compile time. For a datetime value, we check if the MS part is zero then adjust the precision to 0 (min fsp) or 6 (max fsp). No respect to compile time type at all.

Honestly I can file 10+ bugs easily about the above issues.

What is changed and how it works?

Don't get scared of this huge PR - 90% are tests and 80% of which are generated, the core is quite simple. This PR includes the following changes:

1. The core change is in expression/builtin_time.go:
   1. Refine the type inference about the return type and precision of date_add and date_sub.
   2. Refine the evaluation so that the type and precision by type inference is respected.
   3. Add necessary signatures - ideally we should be able to leverage implicit cast to get a datetime value from int/double/decimal/string, however, as described in this comment, to comply with MySQL's weird behaviors we need specialized overload for each of int/double/decimal/string.
   4. Abstract things like "add or sub", "how to evaluate the datetime value", and "how to evaluate the interval value". Use some simple composition to eliminate the mostly-copy-paste individual overloads. And the overall LOC is reduced from 1456 to 591.
2. The vectorized versions are simple enough to not use generator, thus moved from generator (expression/generator/time_vec.go) to regular implementations (expression/builtin_time_vec.go). I would have to write more code in template than just directly write them in go. However the vectorized tests are enhanced by adding more cases.
3. Tests:
   1. Add thorough type infer test cases.
   2. Update several existing UT and integration test cases that already mismatch MySQL's result - each of them is a bug-to-be-reported.
   3. Add more cases to existing UT and integration test.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

Fix the issue that the return type and precision of function `date_add` and `date_sub` does not match that of MySQL.

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has not been approved.

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Details

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[ti-chi-bot]

@ti-srebot: This cherry pick PR is for a release branch and has not yet been approved by release team.
Adding the do-not-merge/cherry-pick-not-approved label.

To merge this cherry pick, it must first be approved by the collaborators.

AFTER it has been approved by collaborators, please ping the release team in a comment to request a cherry pick review.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[ti-srebot]

/run-all-tests

[ti-srebot]

@zanmato1984 you're already a collaborator in bot's repo.