Generator functions in Textfile subst expansion are being called with for_signature=True

[acmorrow]

Describe the bug
The following program used to write val to target, but with current HEAD, it now writes sig:

env = Environment()

def generator(source, target, env, for_signature):
    if for_signature:
        return "sig"
    return "val"

env['GENERATOR'] = generator

env.Textfile(
    target="target",
    source=[
        "@generated@",
    ],
    SUBST_DICT={
        '@generated@' : '$GENERATOR',
    },
)

Required information

• Link to SCons Users thread discussing your issue.
  Discussed on discord with @bdbaddog and @mwichmann.

• Version of SCons
  HEAD

• Version of Python
  N/A

• Which python distribution if applicable (python.org, cygwin, anaconda, macports, brew,etc)
  N/A

• How you installed SCons
  Clone of master

• What Platform are you on? (Linux/Windows and which version)
  Linux, but I'm pretty sure this is platform independent

• How to reproduce your issue? Please include a small self contained reproducer. Likely a SConstruct should do for most issues.
  See the above repo. With SCons 4.2 and older the file target will be produced and contain val, but on HEAD at f6cefb2 it now contains sig. This broke one of the MongoDB custom tools.

• How you invoke scons (The command line you're using "scons --flags some_arguments")
  You can run the above reproducer just by running scons.

[bdbaddog]

Likely this is coming to light due to pr #4021 where env.subst() is now called with env.subst(value, raw=1) to avoid incorrectly subst'ing $$(

[mwichmann]

Okay, after getting a headache going through needless recursion, it's pretty clear where this comes from: in StringSubber's expand method:

                s = s(target=lvars['TARGETS'],
                     source=lvars['SOURCES'],
                     env=self.env,
                     for_signature=(self.mode != SUBST_CMD))

The mode comes straight down from how we were called into this chain - the value of raw to the env.subst call (which is made by the Textfile action) is eventually saved as self.mode. My tenous grasp of logic suggests that last kwarg should maybe be for_signature=(self.mode == SUBST_SIG). Wonder how much stuff that would break...

[mwichmann]

surprisingly, it breaks nothing in the testsuite, and fixes the case in this report.

[bdbaddog]

Yup. Seems like the right solution to me!
That codes from 2005.. :)