retire radioserver: the emulators hold the chip now

[A13xB0]

This repository builds MeshCore for this host. radioserver was here only
because the chip model used to be here too: it wrapped VirtualSX1262 behind a
socket so QEMU and Renode could reach the same model a native node links. The
model moved to MeshBench/virtual-sx1262
in #14, and both emulators now load it directly, so nothing calls this and
nothing needs to.

Why the socket was worth removing, beyond the round trips

• It framed SPI a byte at a time, a path less exercised than the buffered
  one. That is how the model came to answer GetRssiInst correctly for a native
  node and with zero for an emulated one for months, which is where the firmware
  takes its entropy: every emulated board derived its identity from a constant.
• DIO1 could only be polled, because the protocol was request-response. A
  millisecond is a long time to hold a receive interrupt, and a packet delivered
  and acknowledged inside one sampling gap raised the pin never.
• It put three clocks in three processes in front of anybody asking what
  happened when.

What stays

The submodule. A native node compiles the model in, which is what keeps it the
same chip the emulators load, from the same commit. NOTICE.md now says that
rather than the old reason.

The published radioserver-v1..v3 releases stay where they are, so an older
MeshBench keeps working. Nothing new is cut.

What goes

bridge/radioserver.cpp, the radioserver branch of build.sh, its two
release jobs, its pull-request job, RADIOSERVER_PLATFORMS, the
radioserver_only input, and radioserver.version.

Checked

simple_repeater for linux-amd64 still builds from this tree, and the
pull-request workflow that remains is the job that proves it.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Q9HbD44EKWWTRYgxbFGxf6