Glasgow Embedded / Glasgow - History

software: use today's trixie snapshot to build firmware.

This allows us to use sdcc 4.5.

firmware: make commit hash embedding more robust and explain it.

support.usb.libusb1: don't crash if hotplug is unsupported.

software/deploy-firmware: skip group addition if group already exists

Running the script on a distribution where the current user is using gid
100 fails, as that group already exists inside the image:

```
Setting up libgssapi-krb5-2:arm64 (1.20.1-2+deb12u3) ...
Setting up libgdbm-compat4:arm64 (1.23-3) ...
Setting up libperl5.36:arm64 (5.36.0-7+deb12u2) ...
Setting up libcurl3-gnutls:arm64 (7.88.1-10+deb12u12) ...
Setting up perl (5.36.0-7+deb12u2) ...
Setting up liberror-perl (0.17029-2) ...
Setting up git (1:2.39.5-0+deb12u2) ...
Processing triggers for libc-bin (2.36-9+deb12u10) ...
+ groupadd --gid 100 caller
groupadd: GID '100' already exists
```
(nonzero exit code, aborted build)

It's fine to skip group creation in this case, we care about the group
to exist, not how it's called.

manual: revisions: add legend diagram from 1b2 (#418).

pyproject.toml: fix spi-flashrom module name

The module was temporarily called flashrom instead of spi_flashrom, but
this isn't the case anymore.

pyproject: constrain python-version in a less ambiguous way

`uv` pointed this out to me while running glasgow; the `~=` operator
for dependency resolution is a bit subtle; it basically allows you to
"round up" the upper bound of a dependency but it does so based on the
least-sigificant-specified component of a version number.

Basically that means `~=3.11` is equal to "from any version 3.11 up-to
< 4.0" but `~3.11.0` means "from any version 3.11.0 up-to < 3.12.0".

We want the first, but it's apparently subtle enough that uv (probably
correctly, tbh) warns about it. Let's write it in a more unambiguous
way.

Signed-off-by: Austin Seipp <aseipp@pobox.com>