...just to have it handy when it's in need next time.

"-a arch" is not requisite either; and having bunches
of empty lines in the resulting pkglists that are user
visible at least within the conventional installer's
alterator-pkg (groups selection) module wouldn't be nice.

I chose to sacrifice empty-line separators for clarity;
the really good cleanup would save *single* empty lines
between chunks of non-empty ones (not at the pkglist's
start or end); feel free to implement that as well.

This has been clearly lacking while making the previous commit
but the implementation isn't that clear so let it be a separate
step.

The problem requiring the change in subsequent processors
is that these relied upon "@arch" as a flag to be inspected,
and "pkg@!arch1,arch2" on arch2 needs to take out *all* of that
fragment *including* arch1 mention as well.

Part of the cause is difference in handling: "positive" multi-match
would explode its "client" line into multiple lines to filter down
the pipeline, while "negative" multi-match *has* to keep that line
on a similarly single line (otherwise we'd end up with N-1 of those
slipping past the filter for particular architecture thus defeating
the whole purpose of "negative" matching semantics):

$ echo 'pkg@!E2K,mipsel,riscv64' |
  sed -r  ':loop; s/^((([^@]+@!)[^,]+)+),([a-zA-Z0-9_]+)/\1@!\4/; t loop'
pkg@!E2K@!mipsel@!riscv64

I've tried my best to test this specific change but it still might
introduce a regression in some corner case; feel free to report;
looks like there's a space for improvement in m-p's automated
tests department as well.

So now we can do:

  pkg@!ARCHES1,ARCHES2,arch3,arch4

and have pkg excluded on arches mentioned; the previous approach
could only offer explicit whitelists (not that it was entirely
wrong but then again, we have both ExclusiveArch and ExcludeArch
rpmtags in our spec files).

This has been inspired by a few commits that cared
for package availability reasons on a particular
architecture; the problem at hand is that pkglists
might need to include groups of packages that are
(un)available on groups of arches, and tackling that
with plain pkg@arch just results in combinatorial
explosion of that matrix.

Arches are handled one-by-one with a few hardcoded
macro substitutions.

Exploding a "pkg@arch1,arch2" string into a set of:

pkg@arch1
pkg@arch2

with subsequent archdep pruning would do the trick;
so here's another sed oneliner that does just that:

$ echo 'pkg@X86,ARM,ppc64le' |
  sed -r ':loop; s/^((([^@]+@)[^,]+)+),([^,]+)/\1\n\3\4/; t loop'
pkg@X86
pkg@ARM
pkg@ppc64le

See-also: 9601a9e7
See-also: 5581dc91
See-also: http://stackoverflow.com/a/55781741/561921

There are use cases when having this common knob would be desirable.

This code started breeding within pkg.in/*/Makefile,
and it was lacking both E2K (as a generic alias,
not just "e2k or e2kv4") _and_ the ability to negate
selection, as in "everywhere but not on any e2k").

Let's fix all of that at once, and parallelize sed
execution as well; my tests with 1000-line file
containing "a b c d@!E2K e@e2k f g@!X86 h i@IA32"
lines show roughly 3x higher CPU load and lower
execution time (~0.3 sec vs ~0.9 sec on 801-PC).

And turn that code snippet into a proper filter
with inline edit capability (I've dropped the exit
trap as failing to mv signals a disaster anyway).