Re-organization of crates

[mathias-arm]

This PR moves all Veracruz crates in crates/ directory (except for ones under workspaces).

[mathias-arm]

Will rebase after #651 is merged.

[egrimley-arm]

I think the reasons for doing this should be recorded in the PR or in the log message. (Was there something about making grep easier?)

If I've understood correctly, this increases the number of symbolic links from 7 to 8 (by adding crates/third-party -> ../third-party) but gets rid of the potential recursion because we now have workspaces/*/crates -> ../../crates instead of workspaces/*/crates -> ../...

What disruption might be caused by moving third-party into the crates/ directory instead of adding that symbolic link? I can imagine that there might be problems with merging or rebasing when submodules move around, but I can't remember trying it so perhaps it just works?

On second thoughts, perhaps it's better to keep third-party/ at the top level anyway. But then I have to ask myself why QCBOR and t_cose are not in third-party/.

[mathias-arm]

I think the reasons for doing this should be recorded in the PR or in the log message. (Was there something about making grep easier?)

Yes, I wanted to put the Veracruz Rust code in one place. Having the different crates at the top-level with others directories (in particular workspaces) is making things harder to grep for things.

What disruption might be caused by moving third-party into the crates/ directory instead of adding that symbolic link? I can imagine that there might be problems with merging or rebasing when submodules move around, but I can't remember trying it so perhaps it just works?

Initially I was thinking about moving third-party as well, but moving submodules adds a little bit of friction: when switching between branches before and after the move, git does not clean-up. Not a show-stopper though.

On second thoughts, perhaps it's better to keep third-party/ at the top level anyway.

I am now leaning also in this direction since it does help separating our code from third-party.

But then I have to ask myself why QCBOR and t_cose are not in third-party/.

That could also be addressed (later?).