This adds initial support for writing new/modified maps, which are made up of layer groups (LGB) and terrain (TERA). Neither one is really tested yet, but I started checking some simple examples and it seems to work.
I'm an idiot, these row ids are here for a reason and are important for actually
reading Excel sheets. The API has changed to a read_row() function that takes a
row id, and I'll improve it before release.
I guessed the name, it's more like a collection of layers. This actually fixes a
bunch of mistakes (on my part) as I copied the layout from Lumina, but added a
few changes to make it work with modern zones. There's a whole lot more work
here to do, but now it works to parse New & Old Gridania.
This is to make way for another dat module (for the stuff under the user
folder.) This kind of re-organization was inevitable anyway, and I gave the
structs new SqPack-y names to fit their new home.
We had a few odd dependencies that caused nothing but pain in dependent projects
like libphysis. One of these was libunshield (a C library) that our game_install
feature used, but to be honest this was the wrong library to put this code. It
was really only ever used by Astra, and should live there instead - there's no
reason to have it shared between applications (and it's small enough to be
copied if *you* need it.) Also that also killed the system-deps dependency which
had a significant impact on our build time.
Another dependency was replaced: libz-sys. This is replaced by the pure Rust
libz-rs (through libz-rs-sys) which should simplify deploying physis without
having to worry about manually linking libz or other nonsense. Some leftover
copied code from flate2 can also be removed.
I also removed the visual_data feature as Astra ended up using it anyway, and
the distinction doesn't make much sense now. It was previously to gate some
dependencies needed for visual data extraction, but the bitflags and half crates
are small. I can look into splitting the crate up into more features if needed
later.
A dependency that was erroneously included in the refactoring was quote, which
has been removed. Also ran cargo fmt, clippy too.
This recently tripped up cargo-deny because it depends on paste. I looked at it
and the parts that we need (a few BCn decoding functions) has only one place
where it's used and can easily be replaced.
