Merging ts.el into Emacs core

Following up https://github.com/alphapapa/org-ql/issues/409

@alphapapa listed the following reasons why ts.el may be beneficial compared to the existing time API:

> One purpose of the `ts` struct is to provide convenient accessors for elements of the timestamp without having
> to decode and format the time value repeatedly; whether that's an important enough benefit to keep may be
> debatable, but I did measure a significant performance benefit when I was writing it (and whether user code
> typically uses that pattern is, again, another matter).
>
> The main purpose of `ts` is to provide a unified API. Emacs's default time/date functions are spread across
> various libraries and symbol prefixes, and various functions take different arguments and types, which becomes
> bewildering. The API's using a single data type is a significant benefit, as all of the functions can be
> composed, chained, etc.

- [ ] For performance, it would be nice to provide actual benchmarks with numbers that we can present to Emacs devs.
- [ ] For "unified API", it would be nice to provide examples demonstrating how the new API is better.