Centralize media format classification

[amaztony]

Summary

• add a single service for the media extensions FerrumPix currently displays
• reuse the same video classification in gallery items, previews, export filtering, and upload pickers
• centralize the existing upload MIME type mapping

Motivation

The same media categories were maintained independently by the gallery, viewer, video thumbnail service, and other call sites. Those copies could drift and classify the same file differently.

This PR keeps the currently supported extension sets unchanged. It does not add decoding support, bundle codecs or native libraries, or alter the platform dependency and packaging policies.

Validation

• dotnet restore FerrumPix.sln
• dotnet build FerrumPix.sln --no-restore
• build completed with 0 warnings and 0 errors on macOS arm64

[amaztony]

I’m closing this draft PR for now.

The media-format classification refactor is broader than a local bug fix. It changes a shared classification seam used by the gallery, viewer, previews, export filtering, and upload pickers.

Without an agreed upstream direction for this shared layer, continuing would introduce architectural coupling and additional maintenance cost before the project is ready for it. I would prefer to pause this work and revisit it when the project’s architecture and media-format policy are clearer.

This is not a rejection of the idea itself; it is a decision to avoid adding a cross-cutting refactor without a confirmed project-level direction.