cmd: Add 'index remove' command

[ahmetb]

Fixes #550
Related issue: #545, #483

[ahmetb]

/assign @corneliusweig
cc: @chriskim06 to review as well.

[corneliusweig]

Overall this looks quite good. There are some smaller nits.

And I'd like to discuss how we should treat the --force option. Its very existence is proof that this is supported behavior. If anything, I would prefer talking about "not recommended". Anyway, we should think hard if we really want that. My opinion is -- either support it -- or don't. Right now, we are going both ways which will create problems, because users will make use of --force get lost.
So either we hide --force for the time being, or we make sure that everything else works just fine, if additional plugins are installed. (This will also help us plugin reviewers, just saying :) )

[ahmetb]

Its very existence is proof that this is supported behavior. If anything, I would prefer talking about "not recommended". Anyway, we should think hard if we really want that. My opinion is -- either support it -- or don't.

First of all we expect >95% users won't use krew index cmds. So anything we do here is "advanced" land. Furthermore --force says "not recommended" –which means this is even more advanced.

Here's my primary rationale for having --force: Assume you just added Jim’s index –but Jim messed up one of his manifest files, its parsing fails. So, naturally most our primary cmds (install, update, search) start failing. But you have plugins from Jim, so you can just go and "nuke" that index with the consequences.

--force puts you in an "unsupported" land because you might install plugin foo from indexA , then remove --force indexA, install indexB which also has a foo.

[corneliusweig]

Yeah, that all makes sense. I'm also ok with having --force, but even if we don't recommend using it, we should make sure that the core workflow continues to work without hiccups (install, upgrade, update, list, uninstall). I'm mostly concerned about krew upgrade which bails out when finding an unknown plugin at the moment.

So in a way we do support the core workflow after rm --force even though we claim to not support it. Or is that just dumb?

[ahmetb]

I'm mostly concerned about krew upgrade which bails out when finding an unknown plugin at the moment.

I think there's no reason for us to support all those weird corner cases at the moment. We'll make krew update and krew upgrade tolerant to broken/unreachable indexes –but anything beyond that to make this bulletproof adds SO much complexity and very little in return.

[corneliusweig]

/lgtm
/approve
Nice work!

[corneliusweig]

Suggested change

	option is used ( not recommended).`,
	option is used (not recommended).`,

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ahmetb, corneliusweig

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [ahmetb,corneliusweig]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment