Fix process substitution combined with redirection

[JohnoKing]

The code for handling process substitution with redirection was never being run because IORAW is usually set when IOPROCSUB is set. This pull request fixes the problem by moving the required code out of the !IORAW if statement. The following command now prints 'good' instead of writing 'ok' to a bizarre file:

$ ksh -c 'echo ok > >(sed s/ok/good/); wait'
good

This pull request also fixes a bug that caused the process ID of the asynchronous process to print when the shell was in interactive mode. The fix is to temporarily turn off the shell's interactive state while in a process substitution. The following command no longer prints a process ID, behaving like in Bash and zsh:

$ echo >(/bin/true)
/dev/fd/5

[McDutchie]

Very nice work, thanks for that! So, this will be the first ksh93 version with that whopper of a bug fixed.

There are a few things with the regression tests, though. Here's a patch. Summary:

• These tests seem better placed in io.sh along with other I/O-related stuff, so I moved them.
• I prefer to use backticks in Markdown and 'normal single quotes' in plain text.
• The second test depended on an unportable /bin/true path (e.g. the Mac has /usr/bin/true). The test still works when using the true built-in.
• Variable assignments don't require quoting, as that is a scalar context in which split/glob are never performed.
• OTOH, several expansions that should be quoted were left unquoted, including $tmp which is based on $TMPDIR which could well contain spaces. (I'm aware there are loads more unsafely unquoted expansions in the pre-existing tests as well as other scripts. I intend to either fix all those at some point, or just run the whole thing with split and glob globally disabled as they're barely actually used.)
• >> /dev/null looks like a workaround for some broken shells (such as NetBSD sh before 9.0) that can't do >/dev/null when set -C/set -o noclobber is active. POSIXly, device files are exempt from noclobber, and so they are in ksh. So just use >/dev/null, because if that ever breaks, we want it to fail.
• It now prints the expected and actual values on failure for easier debugging.
• I reactivated and tweaked the pretty-printing of times output in shtests and made it subject to a bug/feature test for compatibility with old ksh versions.

[JohnoKing]

The pull request has been updated with your patch. Thanks for pointing out the times pretty-printing in shtests, I missed that by accident.

[McDutchie]

Almost done. Current version removes an empty line in basic.sh for no apparent reason.

edit: which was caused by my own patch, sorry about that.

[JohnoKing]

The pull request no longer changes basic.sh.