Improve file download capabilities

[perryr16]

Any background context you want to provide?

Files with commas in their names (e.g., 12,props,eui,80.xlsx) were downloading as zero bytes from the dataset detail page. This only affected ImportFiles uploaded through the dataset upload endpoint.

For context, there are 6 'file' models that can be downloaded: ImportFile, BuildingFile, InventoryDocument, AnalysisOutputFile, AnalysisInputFile, and ResultFile. Of those, AnalysisOutputFile, AnalysisInputFile, and ResultFile are named internally and are not at risk of having an invalid name.

Downloading InventoryDocuments, BuildingFiles, and ImportFiles from the inventory detail page succeeded even if those files contained commas because Django's FileField automatically sanitizes filenames on upload.

What's this PR do?

Updates the media retrieve endpoint to serve directly through Django instead of nginx X-Accel-Redirect which cannot handle commas or special characters in the filename:

• Simpler and more reliable code
• A few hundred milliseconds slower for very large excel files, otherwise the delay should not be noticeable
• Backwards compatible: should work with all existing ImportFiles

To prevent future issues, imported files are now sanitized on upload to remove problematic characters.

How should this be manually tested?

• Upload an import file with commas in the filename, attempt to download it from the dataset detail (/data/<id>/).
• For further testing, test ImportFile, InventoryDocument, and BuildingFile downloads from the inventory detail (/properties/<id>/) page. These downloads should already succeed on the develop branch.

What are the relevant tickets?

#5075

Screenshots (if appropriate)