osmo-bts-sysmo L1: unify hLayer3 handling
In https://gerrit.osmocom.org/264 , some hLayer3 use was added.
However, some of the messages were already using hLayer3.
After the patch, the prim_init() sets a hLayer3 which is overwritten
shortly after that, so the other functionality is not affected by this patch.
Some functions to generate hLayer3 for a timeslot and similar already existed,
and also to retrieve a timeslot and similar back from a hLayer3 handle.
And also, the other hLayer3 functions use a "magic" byte
(like the most significant byte always set to 0xbb).
So in fact, we should have rejected this patch, and we should follow up with
a merger of my hLayer3 to the existing hLayer3 infrastructure.
The priority is low though, since no harm is done.