Make shared formula cell replacement more greedy

[OSjoerdWie]

Fixes the issue described in #663

Note that I made an additional (non issue related) commit due to build failure of the forked repo.

[jmcnamara]

The core change looks good. I have made some change requests to other parts of the PR.

@louisdewar can you double check this. The change came in with your commit 38ef0e3

Note that I made an additional (non issue related) commit due to build failure of the forked repo.

I'm not sure why that is. It didn't fail on main. I'll look into it. But either way, omit it from this PR.

[jmcnamara]

Could you:

1. Mark the comments as resolved, if they are resolved.
2. Fix the formatting issue.: cargo fmt -- --check
3. Squash the update into the main commit.

[louisdewar]

Hi, yes I wasn't thinking about defined names with this.
This fix seems generally good, as far as I can tell the logic is to expand the definition of what is considered to "possibly" be a range to include characters that constitute defined names and then let the defined name get rejected later on when it's seen that it's not really a valid range. Before it would tokenise and think that was a valid range, now it would tokenise <NO1_VARIABLE> and reject it as a range to be "replaced".

I think there's a few issues with some edge cases:

1. NO1.VARIABLE
2. \A1

I believe these are all "valid" defined names which I think will get treated as ranges.

I think you can fix by just adding '.' and '\' to the char compare.