Safe a couple of needless calls to vslc_vsm_check when we know that
there are no new records and this no need to check shm references /
the next cursor.

vsl sequence and segment updates didn't happen atomically, so
vslc_vsm_check could report spurious overruns.

Replace sequence and segment index with a counter (segment_n), which
always increments (with overflow after UINT_MAX). The actual segment
index will be segment_n % VSL_SEGMENTS. Overrun detection by
calculation of the difference between two segment numbers becomes
simple and safe because we only ever access/update a single integer.

Update the shared memory log head.

(struct VSLC_ptr).priv is now the the equivalent of segment_n from the
reader side. It gets initialized once and is maintained independently.

Patch prepared in collaboration with Martin Blix Grydeland
<martin@varnish-software.com>

Fixes: #1747

properly send any cached req.body along (this didn't work before).

This also makes struct vbc an entirely internal affair in cache_backend,
as it (probably) should be.

the HTTP1 code.

does not pollute the http1 code.

rid of it entirely.

stolen.  This should make it less tempting to do illadvised things.

This is always followed by VCL_call MISS

is still engaged with it.

Simplify the "do a single retry" logic slightly while here.

Add more asserts.

This probably fixes #1675.

Initial diagnosis by:	martin

Submitted by:       controversy187 via github

If you add a probe named "default" all backends will get it automatically,
unless they have their own .probe configuration.

create a backend, that may be true of VCC compiled backends, but
won't be for VMOD dynamic backends.

Eliminate some excessive copying at the same time.

discard them with the VCL.

busy count to drop to zero.

(ie: the VCL is warm)

Duplicate b00043.vtc was no fun.

Lacking a better idea, I'm moving all local tests up 10000
units, so our local test case is now b10043.vtc.