sequtils.nim: Change some func back to proc

[ee7]

This PR is a partial revert of commit 6f57eba (#16293), in case that's preferred to a full revert.

It changes the funcs that take a proc parameter back to procs.

See: #16303 and #16304 (comment):

To confirm: do we want to change a proc that has a proc parameter into a func?

No, it was an oversight by me. We don't want that. Callbacks can have effects.

Note that this PR doesn't fix regressions that @timotheecour mentions in #16305. For example:

import std/sequtils
type Foo = ref object
  x: int
let a1 = zip(@[1,2], @[1,2]) # ok
let a2 = zip(@[Foo(x: 1)], @[Foo(x: 2)]) # error

still produces this error when compiling with --experimental:strictFuncs:

/Nim/lib/pure/collections/sequtils.nim(248, 8) Error: 'zip' can have side effects
an object reachable from 's1' is potentially mutated
/Nim/lib/pure/collections/sequtils.nim(284, 17) the mutation is here
/Nim/lib/pure/collections/sequtils.nim(284, 17) is the statement that connected the mutation to the parameter

[ringabout]

I think adding module to tests/effects/tstrict_funcs.nim isn't enough to cover all functionse. It needs more care to change them. Maybe add --experimental:strictFuncs matrix to the single test of this module(testament).

[ee7]

CI failure is unrelated:

   Compiling /home/runner/work/Nim/Nim/pkgstemp/fidget/tests/test (from package fidget) using c backend
  /home/runner/.nimble/pkgs/typography-0.6.0/typography/font.nim(25, 22) Error: undeclared identifier: 'Segment'

Reported here treeform/fidget#121

[timotheecour]

No, it was an oversight by me. We don't want that. Callbacks can have effects.

There's a disagreement here, so maybe this revert should wait until this is clarified. The current nim implementation allows func(cb: proc(): int): int, and I'm arguing the current nim implementation is the correct, useful, safe behavior, see #16303 (comment)

[ee7]

There's a disagreement here, so maybe this revert should wait until this is clarified.

Sure. I just made this PR so it's easily available if it's what we want.

I think adding module to tests/effects/tstrict_funcs.nim isn't enough to cover all functions

I was thinking this too. It's true, given that CI is green despite devel having a strictFuncs regression with e.g. zip and ref types.

[ee7]

Now that fidget was fixed, could somebody restart the two failed package CI jobs? I don't have the permissions to do it.

[ee7]

Rebased on devel to resolve a merge conflict due to 8f6e07a.

The relevant diff is only:

 template applyIt*(varSeq, op: untyped) =
-  ## Convenience template around the mutable ``apply`` proc to reduce typing.
+  ## Convenience template around the mutable `apply` proc to reduce typing.