Naming sanitization leaves double spaces, trailing non-breaking spaces, and glues elided words (apostrophes)

[Mika3578]

Describe the bug

The naming/sanitization helpers used when building file and folder names produce malformed names in several cases. All of them come from sanitize_filename() (medusa/helper/common.py#L295) and sanitize_scene_name() (medusa/helpers/init.py#L748), which is used by the %S.N / %E.N tokens via dot() in medusa/tv/episode.py.

1. Double spaces after removing forbidden characters. sanitize_filename() deletes :"<>|? without collapsing the surrounding whitespace. A show named Example Show : Subtitle becomes Example Show Subtitle (2016) (double space). A name ending in ? can leave a double space before an appended year.

2. Non-breaking space (U+00A0) is never handled. It is not converted to a separator by sanitize_scene_name(), not removed by sanitize_filename(), and not stripped by the final strip(' .'). Episode titles from TheTVDB can contain NBSP. Result: filenames that end with an invisible non-breaking space, which breaks later path comparisons. Tabs were handled in Strip more invalid filename characters - Fixes #3839 #4351; NBSP remains.

3. Deleted apostrophes glue words together. sanitize_scene_name() deletes ' and ' outright. Fine for English scene style (Bob's Burgers -> Bobs.Burgers), but for elisions it merges two different words: L'exemple -> Lexemple, C'est un titre -> Cest.un.titre.

4. Minor, same code path: an empty episode title yields a double dot (Show.S01E01..(2025-01-15)), and %E_N/%S_N leave trailing underscores when a title ends in ?/! because us() converts spaces before punctuation is removed.

To Reproduce

from medusa.helper.common import sanitize_filename
from medusa.helpers import sanitize_scene_name

sanitize_filename('Example Show : Subtitle')   # 'Example Show  Subtitle' (double space)
sanitize_filename('example title\xa0?')          # 'example title\xa0' (trailing NBSP)
sanitize_scene_name("L'exemple de titre")      # 'Lexemple.de.titre' (glued words)

Or add a show whose name contains : and let Medusa create the show folder.

Expected behavior

• Removing a forbidden character should not leave double separators; e.g. Sonarr collapses repeated separators with ([- ._])(\1)+ and uses a "smart" colon replacement (": " -> " - ", remaining : -> -) in FileNameBuilder.CleanFileName.
• NBSP should be treated like a regular space (or stripped at the ends like a space).
• Apostrophe removal in the naming context should insert a separator instead of merging words (search/scene matching in name_cache.py / scene_exceptions.py must keep the current behavior, so the change belongs in the dot() naming helper, not in the shared sanitize_scene_name()).

Medusa (please complete the following information):

• OS: Windows 10 (also reproducible on any OS; code path is platform-independent)
• Branch: develop
• Commit: b935f77 (current develop; both functions unchanged)
• Python version: 3.13
• Database version: n/a (not DB related)

Additional context

Related history: #3839 / #4351 added \t and control characters to sanitize_filename(); NBSP was the remaining invisible character.

Edited to remove personal library show/folder names from the original report.