When we pushed the object allocation into the stevedores for
-spersistent, we did not add LRU eviction to that allocation
path.

Then we added the Transient storage as a fallback for objects
we could not allocate, and they all went there.

Change the way object allocation works as follows:

        If VCL set a stevedore hint and it is valid, we stick
        with it, and LRU that stevedore attempting to make space.

        If no valid hint is given, try all stevedores in turn,
        then LRU one of them to make space.

Fixes	#953

an object.

Fixes	#985

are willing to evict per storage allocation attempt.

Previously this was hardcoded at 50, make it 10 instead.

The ABI we give vmods consist of the Varnish version number and the
git commit ID meaning we can break ABI at will.

Output a warning if we can't determine git commit ID

data, the thread would go into a spin.

Fixes	#942

Found & Fixed by:	Martin

Fixes	#978

Found & Fixed by:	Martin

Fixes	#979

	we transform (gzip/gunzip)
	we stream and backend didn't send c-l.

Fixes	#980

Testcase by:	Martin

If the first (policy-chosen) backend fails to get us a connection,
retry a random backend (still according to their weight) until
retries are exhausted.

Kristian sent a proof of concept patch, I just cleaned it up and made
it compile.

Thanks to:	Kristian

Fixes	#977

32 bit systems.

Spotted by:	Kristian

Explained the semantics quickly so people understand why not all objects survive a crash

more on transient.
didn't get it rigth the first time.
Thanks to scoof for pointing it out.
(cherry picked from commit 6a59dacccb1fc402c5d1cd703c048835ba400c38
and 0b23a0f083865f75905bf66dde80fc4ac6458463)

response code stuff and fully incorporate the fact that they are both
16 unsigned now.

No functional changes.

the silo into VM.  If this fails or we get garbage, the silo will
be cleared.

Fixes	#962

end of the UINT_MAX items we can support.

Found by adding a lot of asserts and brute force testing, both
of which I have left in.

Fixes       #967

May also be relevant to #827
(cherry picked from commit 726d93ff5a815774d7a3b4a23dcba2efb3c08ca7
and a6ddafc0e87bcec71f016b8cae77b78641e617d7)

also have the 206/range handling.

Instead of building the 304 response from scratch, dumb down the
200 response accordingly.

This implement the policy consideration of #970 led to.

Written by:	Martin

prefer IMS, it is less work and less data to transmit.

Fixes	#972

Thanks to:	Martin

into the RFC policy module where it belongs.

Put esi → do_esi in upgrade checklist, fix markup typo.

Fixes #974

Document the TTL record.

Simplify the move from sp->wrk to sp->obj accordingly.

Add a diagram that explains how ttl, grace & keep related to
each other.

Fix VCL's obj.ttl variable to appear to be relative to "now" rather
than obj->entered. (#956)

Also log SLT_TTL when we change obj.grace and obj.keep

Make SLT_TTL always have the same format:
	XID
	"RFC" or "VCL"
	obj.ttl
	obj.grace
	obj.keep
	obj.entered
	obj.age
In addition "RFC" has:
	obj.date
	obj.expires
	obj.max-age

Fixes	#956

Thanks to:	DocWilco