as agreed on -commit.

Closes #2906

Refs 832d6da2

Fixes #2834

Fixes #2185

Test case by @fgsch

Thank you to phk for the improvement over the initial suggestion

When running really massive runs, "-j180 -n10000" kind of things,
the "rm -rf" of the tmpdir becomes the limiting factor.

The new -C option sends that int nice(1)'ed child process.

Two of them have no init step, and the third one is fail-safe.

And get a chance to catch vdp->fini() failures with the offending VDP
still present in the list of filters.

also when nothing has been received.

but also make it fail faster when appropriate

Apologies for the spam, I didn't think this would be a recurring
pattern.

Fixes: #2844

I caught one during a review and figured I might as well sweep the whole
tree, so this patch was created using Coccinelle and the following steps:

    $ cat >check_obj.cocci <<EOF
    @@
    expression obj, magic;
    @@

    if (obj != NULL) {
    - CHECK_OBJ_NOTNULL(obj, magic);
    + CHECK_OBJ(obj, magic);
    ...
    }
    EOF
    $ spatch --dir . --in-place --sp-file check_obj.cocci

This is my fourth semantic patch, I wouldn't mind checking them in to
have them handy in the repo to semi-automate code polish every now and
then. For example in a /tools/cocci directory. This way people could
study them or use them in their out-of-tree projects like VMODs or VUTs.

Fixes: #2297

Spotted by:	Nils

Tests #2887

Tests #2852

Sorry, not the big overhaul of u6 yet...

Neither the vmod_ default $Prefix not an author-supplied $Prefix has
got anything to do with the actual name of the vmod.

Allow VMOD writers to permit with NULL_OK flag.

Only call object destructor on initialized objects.

Fixes #2839