VRT_CTX used for VCL events.

Acquiring a read lock on the CLI thread is unnecessary since a write
lock may only be acquired on that very thread.

The locking order when both are required is:

    (struct vcl).temp_rwl => vcl_mtx

Fixes #2008

There is currently no locking around temperature changes despite the
fact that it may be read outside of the CLI thread to make decisions.

In the case of a panic, instead of locking we simply make sure not to
perform an incomplete read of the pointer.

Pass childs status on failure

Make atexit more robust

thanks to @dridi for the dictionary lookup

Maybe it's just my brain that is incapable of considering the
fact that a bit of macro-fillin may in fact be #if 0 - disabled
here, so this should hopefully help me (and others?) to not
spend more life time than absolutely necessary on feeling stupid.

if they are signed or unsigned.

Range-check enums is sound programming practice, but that concept
seems to be beyond the imagination of certain compiler people:

   vhp_decode.c:96:2: error: comparison of unsigned expression >= 0 is always true [-Werror=type-limits]

Add a dummy "MIN" value to the enum, give it value -1 to force
the compiler to use signed ints for the enum, and then check
that they're never negative.

Fixes #2084

To avoid another regression, a test case now passes a probe to the debug
VMOD. This problem was discovered while working on #2068.

Also try to clarify CLIS_PARAM (= nonsensical parameter) from
CLIS_CANT (= params make sense, but you can't do that.)

Spotted by: Dridi Boukelmoune

Thanks @nigoroll for the suggestion.

This allows to test vgc_inistep only once and make it easier to work for
the compiler.

change resp.msg to resp.reason

Spotted by: Nils Goroll