feat: expose Period::new by introducing a PeriodUnit enum

[gibbz00]

The Perdiod::years and Period::months may feel redundant now, but keeping them to not break any public API.

We're building an interface at work. Doing this change upstream removes the need for us (and probably others) to create a data type that solely acts as a bridge.

[djc]

It's unclear to me what the benefit of this change might be. This change definitely already breaks the public API, so it doesn't make much sense to retain some parts of it.

Also note that this crate is currently being maintained on a volunteer basis. If you're going to start depending on it for business-critical purposes, I suggest we should have a conversation about commercial support.

[gibbz00]

Did I "break" the public API? My intent was to only extend it. Perhaps our definitions differ 😊

Good point. I'll send you a email later today after having a conversation with my bosses. Where would you prefer us to send it to? dirkjan@ochtman.nl?

I have some fairly uncontroversial PRs lined up that I wish to send in during the meantime. Hope that that's ok by you.

[djc]

Did I "break" the public API? My intent was to only extend it. Perhaps our definitions differ 😊

Ah, sorry -- I was mistaken. Would still be good to discuss why this is an improvement for you.

Good point. I'll send you a email later today after having a conversation with my bosses. Where would you prefer us to send it to? dirkjan@ochtman.nl?

Yup, that should work!

I have some fairly uncontroversial PRs lined up that I wish to send in during the meantime. Hope that that's ok by you.

I'll take a quick look and will decide whether I want to review as a volunteer per PR. 👍

[djc]

FWIW, I don't think "deprecate" is a great description for just killing API. :)

[gibbz00]

Heheh, yeah. Should probably just have written "remove".