Add byte offset functionality to StorageIO and Access API

[landreev]

We want to be able to support the standard HTTP "bytes=" parameter. For example, call /api/access/datafile/<id>?bytes=8192-8208 and get just the 16 bytes at the 8K offset, instead of the entire file.

(it's "working" for some people in our own prod. Dataverse - see #5957 - but only because we redirect to S3, which does understand the parameter).

To implement it, we should add some form of "open at offset specified" functionality to the StorageIO structure. Then adding the handling of the extra parameter to the Access API should be trivial.

Should be straightforward overall.

[djbrooke]

• Small modification of AccessIO
• The benefits are potential cost savings/stability for installations not running on S3, streamlining file integrity checks on direct uploads to S3 for installation running on S3, and better support multi-part download through WGet-continue

[djbrooke]

• from @qqmyers - this will allow us to do better file ID for large files
• this will continue to support the Tectonic use case on Harvard Dataverse - and we're concerned about that same use case happening for another installation, where there'd be cost/stability implications