bug#62031: 29.0.60; python-mode indentation after re.match

[Lele Gaifax <lele@metapensiero.it>]

I had a quick glance at this, and while I can confirm the problem with
the following test

    diff --git a/test/lisp/progmodes/python-tests.el b/test/lisp/progmodes/python-tests.el
    index 4f24c042c6a..9926c4b002f 100644
    --- a/test/lisp/progmodes/python-tests.el
    +++ b/test/lisp/progmodes/python-tests.el
    @@ -1982,6 +1982,17 @@ python-virt-bin
        (should (eq (car (python-indent-context)) :after-block-start))
        (should (= (python-indent-calculate-indentation) 8))))

    +(ert-deftest python-indent-after-re-match ()
    +  "Test BUG 62031 regression."
    +  (python-tests-with-temp-buffer
    +   "
    +def test_re(string):
    +    if re.match('^[a-c]+$', string):
    +        print('yes')
    +    else:
    +"
    +   (python-tests-look-at "else:")
    +   (should (= (python-indent-calculate-indentation) 4))))

I could not figure out how this can be fixed, with my current knowledge
on python.el.

There are several places where the regexp produced by `(python-rx
block-start)' is used, many within a `(looking-at ...)' form and several
others in a `(re-search-forward ...)': given that block starting
"keywords" are required to be at the beginning of a line, possibly
preceded by whitespace, I think that the problem may arise from those
usages which do not explicitly enforce the constraint, but unfortunately
all my attempts to do that a) didn't make the test above green and b)
broke one or more of the other tests.

I will try again, but I'm afraid this will go beyond my current spare
time availability.

bye, lele.