Fix UnicodeDecodeError when filtering a binary POST body (#660)

[uttam12331]

Fixes #660

Problem

When filter_post_data_parameters is configured and a request has a raw binary POST body passed as bytes (e.g. an image/file upload, rather than a BytesIO), replace_post_data_parameters crashes:

from vcr.request import Request
from vcr.filters import replace_post_data_parameters

body = b"\xff\xd8\xff\xe1\x00\x08Exif=\x00\x00MM*"   # raw JPEG-ish bytes
req = Request("POST", "http://x/upload", body, {"Content-Type": "image/jpeg"})
replace_post_data_parameters(req, [("session", "REDACTED")])
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0

The else branch treats any non-dict / non-JSON body as form-urlencoded, splits on b"&" / b"=", and does k.decode("utf-8"). Any 0x3D (=) byte inside binary data makes the preceding binary run be parsed as a form "key", so the decode blows up. The existing isinstance(request.body, BytesIO) guard doesn't help because a raw bytes body isn't a BytesIO.

Fix

Wrap the form-parameter parsing in a try/except UnicodeDecodeError. A body that isn't valid UTF-8 form data has no form parameters to replace, so it's left untouched instead of crashing. Text / form-urlencoded bodies behave exactly as before (the new body is only assigned on success, so there's no partial mutation).

Verification

• The reproduction above now leaves the body unchanged instead of raising.
• Normal form filtering is unaffected (b"id=secret&keep=1" → b"keep=1").
• Added test_replace_post_data_parameters_binary_body, which fails on master and passes with this change; the full tests/unit/test_filters.py suite passes (25 passed). ruff clean. Changelog updated.