The .VSC file steals .RST's directive syntax, but it is not a RST
file, and it is not processed as such.  Maybe later it will become
a .RST file, but that is TBD.

A C-struct for use in varnishd is built, (along with a JSON structure
which describes the counters.)

For now that structure is identical with what comes out of
include/tbl/vsc_f_main, so nothing special happens in the VSM yet.

fully understood circular-ish make distcheck dependency problem.

Thank you to @RantyDave

Closes #2326

It's sometimes confusing new users, and it managed to confuse me too a
couple months ago.

Fixes #2329

Spotted by coverity.

Spotted by coverity.

Reported by Coverity:

    ID 1406102:  Incorrect expression  (ASSERT_SIDE_EFFECT)
    /bin/varnishtest/vtc_http2.c: 852 in receive_frame()
    846                         s->wf = 0;
    847                         AZ(pthread_cond_signal(&s->cond));
    848                 }
    849                 continue;
    850         }
    851         AZ(pthread_mutex_unlock(&hp->mtx));
    >>> CID 1406102:  Incorrect expression  (ASSERT_SIDE_EFFECT)
    >>> Argument "vtc_error" of assert() has a side effect because the
    >>> variable is volatile.  The containing function might work
    >>> differently in a non-debug build.
    852         assert(vtc_error || vsb == NULL);
    853         return (NULL);
    854 }
    855
    856 #define STRTOU32(n, ss, p, v, c)                             \
    857         do {                                                 \

Although I'm not sure how relevant it is since assert() from vas.h
doesn't remove side effects when no-op'd.

The name idle_send_timeout does not seem to describe well where the timeout
applies (individual write()/writev() calls for arbitraty amounts of data).

The fact that the explanation is complicated might be an indication that a
different timeout implementation be more appropriate.

HTTP2 should probably implement send_timeout somewhen.