fix sanitation for --strip-ansi

[curious-rabbit]

--strip-ansi=always left two classes of ANSI escape sequence in the output:

• 8-bit C1 introducers (U+0090, U+0098, U+009B, U+009D, U+009E, U+009F). On terminals that interpret 8-bit C1 in UTF-8 (kitty, by default; older xterm and VTE configurations), these are the single-codepoint equivalents of ESC P, ESC X, ESC [, ESC ], ESC ^, ESC _ . They introduce DCS, SOS, CSI, OSC, PM, and APC sequences respectively. Bat's parser treated them as text.
• DCS / SOS / PM / APC bodies. Even when introduced by the 7-bit ESC P/X/^/_, the body up to the string terminator was emitted as text and survived strip_ansi.

This patch teaches EscapeSequenceOffsetsIterator to recognise the 8-bit introducers as their 7-bit equivalents, and to consume string-terminated bodies (via either form of introducer) as a single opaque Unknown segment. Both then drop out of strip_ansi along with the existing CSI/OSC handling.

A small unit test in src/preprocessor.rs covers the three new cases (8-bit CSI, 7-bit DCS body, 8-bit DCS body with 8-bit ST).

Note:
--strip-ansi=always is bypassed entirely when bat selects SimplePrinter (i.e. when stdout is piped or --color=never is set), so bat --strip-ansi=always file | grep … returns the raw escape sequences.
The existing strip_ansi_does_not_affect_simple_printer test seems to lock this in deliberately. is that still the intended scope, or shoudl this be fixed? strip-ansi=always implies it always filters but thats not true

[keith-hall]

Note:
--strip-ansi=always is bypassed entirely when bat selects SimplePrinter (i.e. when stdout is piped or --color=never is set), so bat --strip-ansi=always file | grep … returns the raw escape sequences.
The existing strip_ansi_does_not_affect_simple_printer test seems to lock this in deliberately. is that still the intended scope, or shoudl this be fixed? strip-ansi=always implies it always filters but thats not true

I believe the intention is to leave the output unchanged when piping in scripts (and cat is an alias to bat and strip-ansi comes from the config file instead of the command line...)
Probably bat could benefit from a refactor which would always honor command line arguments, as this is quite a common problem I think, but maybe it makes sense to leave it as it is for this PR, what do you think?

[curious-rabbit]

Closed in favor of #3729