Fix missing CHERRY-PICKING, MERGING, REVERTING, etc

[dahlbyk]

Adding tests for #828

[dahlbyk]

Revert "fix CHERRY-PICKING display bug"

@NihilityT your fix does work, but I ended up reverting it to gain some performance. Not using the status branch name if we have it leads to two extra git calls on every prompt (see failures). Since we nearly always have a branch, we should try to use it as an optimization.

So the logic ends up being:

1. Get branch from status (but only if not in .git or bare)
2. Always call Get-GitBranch to add the necessary suffix, but with new parameters -Branch and -IsDotGitOrBare (which we know from before calling git status)
3. Only make the extra git rev-parse calls for .git/bare if we need to

[rkeithhill]

A few style comments but otherwise LGTM

[rkeithhill]

This is not an exported command but, in general, I'd add new parameters at the end of the parameter list to avoid breaking any callers that use positional parameters. That said, the only caller is in this file - besides to Pester test calls to this command. So, if it makes more sense for the Branch parameter to be first then go for it. In retrospect, we probably should have been more consistent in using PascalCasing on all function parameters given that they correspond to command parameters e.g.: Get-GitBranch -Branch main -GitDir ~\.git.

[dahlbyk]

This is not an exported command but, in general, I'd add new parameters at the end of the parameter list to avoid breaking any callers that use positional parameters. That said, the only caller is in this file - besides to Pester test calls to this command. So, if it makes more sense for the Branch parameter to be first then go for it.

I generally agree, but I wanted to keep $sw last so figured all bets are off.

In retrospect, we probably should have been more consistent in using PascalCasing on all function parameters given that they correspond to command parameters e.g.: Get-GitBranch -Branch main -GitDir ~\.git.

Consistency would be good. Can revisit holistically after I get a release out.

[rkeithhill]

Yes, it would be better to have style changes and semantic changes in separate PRs. Makes them both easier to review. 😁

[rkeithhill]

Huh, I never realized until now that our Inovke-NullCoalescing command handles an infinite list of args. Nice. The new PowerShell ?? operator allows this but not via a single operator e.g. $b = $b ?? $branch ?? { dbg...; git ... HEAD -q } ?? ... And the ?? operator is PS checks for equality to $null and not "truthiness" which can be surprising. And since the PS ?? operator is not in Windows PS 5, we can't use it here.