whisper.cpp backage structure based on llama.cpp hwcaps setup

[Petter Reinholdtsen <pere@hungry.com>]

Dear Christian,

I have tried to rewrite the build setup in
<URL: https://salsa.debian.org/deeplearning-team/whisper.cpp >, but ran
into a problem with how whisper.so.1 is handled.  I suspect I did
something wrong.  I dropped Don-t-build-the-server-example.patch and
cmake-Add-ability-to-pass-in-GGML_BUILD_NUMBER.patch as they did not
seem relevant for whisper, and added 2000-install-binaries.patch to add
install rules for the examples.

Do you have an idea how to get the multi-hwcaps branch working for
whisper.cpp?

-- 
Happy hacking
Petter Reinholdtsen