add `assets_ignore` to include full path in regex search

Is your feature request related to a problem? Please describe.
We structure our assets folders to have different subfolders inside of them, and in a path we have css files we'd like to ignore (used for pdf generation with weasyprint).

so for example:
assets/subpath/base.css

The behavior I'd have expected is to be able to pass ".*\/subpath\/.*" as a valid regex string to ignore all files in the subpath dir. But instead https://github.com/plotly/dash/blob/a7a12d180e16eac0a84b88e2b8dc8f7c7601cbaf/dash/dash.py#L1344 does the regex match only against the file name itself.

I'm curious if others have run into this behavior/issue before.

Describe the solution you'd like
Simplest version is:

files_gen = (x for x in files if not ignore_filter.search("/".join([base, x]) if base else x)

This would obviously be a breaking change for those that prefix scan in their ignore patterns, so might just be me that needs to change. Maybe just a documentation fix that declares that these regex patterns are for the filename and not full path?

Start in dash/dash.py around line 1344, where the ignore regex is matched against asset filenames. Check the surrounding asset-discovery logic and existing tests or documentation to determine the current contract. Done means the requested full-path matching behavior, or an explicit filename-only contract, is covered by tests or documentation without leaving the breaking-change question ambiguous.