Allow reading WACZ from http / https

[wvengen]

file:// and s3:// are supported, but not plain http:// or https:// ... that is an important feature to have in here.

[passionworkeer]

I’d keep the reader path unchanged and only generalize the opener layer: accept file-like objects from fsspec.open/http(s) and stream them into the existing ZIP/WACZ reader. Two practical checks that will prevent regressions: support Range requests so large remote archives do not need full download, and fall back to a local temp copy when the server does not advertise byte-range support. Tests can cover one tiny WACZ served by a local HTTP server plus a no-Range response case.