Basic CI is now implemented, tests are being run on a Raspberry Pi, every time a push happens a run of most test is put into the test queue on the machine. Then any data about failing tests is commented over on GitLab on the last commit in that push.

e.g.
https://gitlab.com/composr-foundation/composr/commit/c53019104dc73210afc655dfaafe820819fa4aa8#note_237467732

(v11 is currently having various tests failing, it'll be cleaner in the future)

I'm leaving this issue open as there is definitely potential for improving this, e.g. having another Pi running a copy of Windows and doing the same. That said, others can implement this on an informal basis and sponsor fixups for their exotic configurations - it does not need to be centralised.

I've opened new issues that cover extended continuous integration, to handle testing of particular maintenance codes we've defined.