Be more efficient with Git

[techman83]

Depending on how we need to do things, rather than deleting the whole directory we could get away with a git reset --hard HEAD (or --hard origin/master) to discard all git changes from the staging area and working directory. That can be followed with git clean -df to remove all untracked files and directories.

In our case that might actually make things more complex, but in theory it should then be possible to just pull down the commit objects between what we had last and what's out there now, which should be faster and smaller than a complete clone each time.

• @pjf

[techman83]

I'm thinking it might be worthwhile making these explicit methods, rather than something that happens on instantiation. It'll also break my brain out of it's current deep recursion.