Replace setup.py with pyproject.toml

[mcoughlin]

This PR tries to replace setup.py with pyproject.toml.

[sahiljhawar]

LGTM. Resolves #241

[sahiljhawar]

@mcoughlin Sorry I didn't notice but all the entry points has been renamed. Should I change all of them?

[mcoughlin]

@sahiljhawar I think by convention, they should be "-" rather than "_"

[sahiljhawar]

yes, but this would mean that old workflow will be broken, those who are accustomed to do light_curve_analysis, etc

[mcoughlin]

@sahiljhawar I think we should make an announcement on list.

[sahiljhawar]

Okay! Docs needs to be updated as well.

[tylerbarna]

@mcoughlin @sahiljhawar once we update the docs, perhaps we should tag this as a new release? since this alters the way users interact with most of the functions

[tylerbarna]

It might also be worth (if possible) having some small period of deprecation where the old names are still callable to minimize compatibility issues with any users who updated without updating their existing codes that rely on nmma

[sahiljhawar]

But if we give the upgrade window, people will still be using the old names. How often do we seriosuly care about numpy or scipy deprecations?👀

[mcoughlin]

I think as long as we do a bump to 0.1.0 we can justify the change.

[tylerbarna]

I think as long as we do a bump to 0.1.0 we can justify the change.

that's a good point

[tylerbarna]

A bump to 0.1.0 may also be a good time to start doing release notes

[mcoughlin]

@sahiljhawar Can you write some release notes and then try a 0.1.0 release? I just pushed changes to the docs.

[tylerbarna]

@mcoughlin @sahiljhawar while we're at it, we may as well remove the KNTimeshift deprecation warning if we're upgrading to 0.1.0

relevant commit: 4e30aad

[sahiljhawar]

@tylerbarna Yes!

[tylerbarna]

@sahiljhawar just submitted #243 to address this