feat: --noise-filter preset + fix binary garbage in parser output

[Vasco0x4]

Summary

Two independent improvements targeting common pain points when running
MANSPIDER against Windows infrastructure.

---

1. --noise-filter {moderate,aggressive} — suppress Windows system noise

Running against domain controllers or file servers produces massive amounts
of results from Windows system paths (WinSxS, PolicyDefinitions, System32…)
that never contain useful data. This adds a --noise-filter flag with two
presets:

Mode	Excluded dirs	Excluded extensions
moderate	PolicyDefinitions, WinSxS, Servicing	.adml .admx .mui .mof .cat .manifest
aggressive	+ System32, SysWOW64, Assembly, Fonts, Spool, Windows Defender	same

Presets feed directly into the existing exclude_dirnames /
exclude_extensions infrastructure, so they compose cleanly with
--exclude-dirnames and --exclude-extensions.

Usage:

manspider <target> -f password --noise-filter moderate
manspider <target> -f password --noise-filter aggressive

---

2. Fix: binary garbage chunks in parser output

Files like PE/DLL/binary were being misidentified as text by
charset-normalizer, producing massive \xef\xbf\xbd garbage dumps in
match output. Fixed with:

• is_text_file() now rejects files where >1% of decoded chars are
  Unicode replacement chars (U+FFFD)
• extract_text() applies the same ratio check after any extraction path
  and falls back to extract_strings_from_binary() when exceeded
• Removed the -a flag from grep to stop binary stdin being treated as
  text (was causing single-line binary dumps even with -m 5)

---

Testing

Tested against an internal AD environment with:

• A DC SYSVOL share (noise-filter moderate/aggressive)
• Binary files (.dll, .exe) in accessible shares (parser fix)

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

0 out of 2 committers have signed the CLA.
❌ @claude
❌ @Vasco0x4
_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}