This was lost in commit 74328de8.
Fixes #2412.

Fixes #2408.

They were inherited from vmod-debug, from a time when vmod-blob didn't
exist in the source tree.

Refs #2362
Refs #2402
Refs #2407

decode table for base64.

Having a stack on the heap just feels unclean, also this way we have a chance
to get a red zone adjacent to the mapping just in case we manage to overflow
the alt stack also.

Ref: #2396

Ref: #2396

No, we do not want to bring back panics for out-of-ws

Ref: #2402

Ref: #2402

My previous patch did not update the manual and the usage message for
varnishtest concerning the default internal buffer size.

Since SunOS needs a litt bit more workspace_thread, this test case
needs to be updated.

The default buffer size for vtc (configurable with -b) had to be
increased.

Fixes #2402.

Closes #2407

a STRING (const char *) to vmod_priv->priv (void *) (const qualifier).

References #2402

The locking around the use of vcl_active does not include the checking
of its magic value or the temperature asserts, leading to a race when
changing vcl_active.

Fixes: #2390

The struct vcl and the temperature rwlock were not free'd on
discarding labels.

We can directly use vmod-purge now.

This module offers a finer-grained control over purges, in the old
pre-vcl 4.0 style. For that, the signatures of VRT_purge and HSH_Purge
were updated to inform about the number of rearmed objects.

Fixes #2403.

This one is for h2 streams, so the thread pools needed to learn about
"stream" tasks. To properly test this counter, thread_pool_timeout is
now disregarded when the vtc_mode debug flag is set. A minimum value
of 10s is unrealistic for the test suite, it is now hard-coded to 0.5s
when ran through varnishtest.