What problem does this PR solve?

Issue Number: close #34998

Problem Summary:

There was once an issue #9716 about extract(day_microsecond) not recognizing the heading day portion, and it was later fixed by #21601. But that PR introduced another problem about extract(day_microsecond) on regular time data. What happens under the hood is:

1. Unit day_microsecond guides the type infer to interpret the time data as datetime through implicit cast;
2. When doing implicit cast from time to datetime, current timestamp is padded to the original time as the date portion of the resulting datetime value;
3. The time portion is messed up with the newly padded date portion, esp. for values larger than 24 hour or negative.

What is changed and how it works?

For regular datetime/timestamp/time types, let extract(day_microsecond/second/minute/hour) interpret it as time.

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 extracting day_microsecond/day_second/day_minute/day_hour from time type emits wrong result.