index: canonicalize directory case when adding

[ethomson]

On case insensitive systems, when given a user-provided path in the
higher-level index addition functions (eg git_index_add_bypath /
git_index_add_frombuffer), examine the index to try to match the
given path to an existing directory.

Various mechanisms can cause the on-disk representation of a folder
to not match the representation in HEAD or the index - for example,
a case changing rename of some file a/file.txt to A/file.txt
will update the paths in the index, but not rename the folder on
disk.

If a user subsequently adds a/other.txt, then this should be stored
in the index as A/other.txt.

For example:

% git init .
% mkdir a
% echo hello > a/file.txt
% git add a
% git commit -mfoo
% git mv -f a/file.txt A/file.txt
% git status
On branch master
Changes to be committed:
  (use "git reset HEAD <file>..." to unstage)

        renamed:    a/file.txt -> A/file.txt

% dir
...snip...
08/04/2015  04:54 PM    <DIR>          a

% echo hi > a/other.txt
% git add a/other.txt
% git status
On branch master
Changes to be committed:
  (use "git reset HEAD <file>..." to unstage)

        new file:   A/other.txt

[ethomson]

Haha, it turns out that I had my last-minute "only run this on core.ignorecase systems" check backwards. Oops! Corrected, I think that this is ready for review.

[carlosmn]

This entry isn't looked up in the assertions below.

[ethomson]

So as you mentioned: there is (was) another mechanism by which we would dup existing entries and keep the original filename as it existed in the index.

I actually think that this is probably not the generally correct behavior: instead, I think that git_index_add should use the given path so that one could perform a case-changing rename. However, I think that git_index_add_bypath and git_index_add_frombuffer should stay friendly to the callers and maintain the path as it exists in the index, if it's already there.

I laid down another commit to update this behavior. Let me know if you don't agree with this plan.

[carlosmn]

I do like that approach. When you ask to insert a specific entry we do that, otherwise we take the case-sensitivity of the worktree into account.

(you seem to have some conflicts, presumably due to the changelog stuff)

[ethomson]

Rebased! I had been waiting on this in case the hash affected anything here. It doesn't, so this should be ready to go.