Write rclone uploads under a temporary name, re-upload short blobs (closes #266)

The rclone backend wrote each object straight to its key, so killing an
upload to a local or sftp remote left a truncated object there. The next
backup found the key, skipped the upload and recorded a snapshot that
could not be restored.

On a remote where rclone says a file can be seen while it is still being
written, an object is now written under a name ending in `.partial` and
moved onto its key with the remote's server-side move, as rclone's own
copy does; listings skip such names. A backup also uploads a blob again
when the stored object's size differs from the blob's.

Judgement call: the temporary name is used only where rclone sets its
PartialUploads feature; other remotes already show an object only once
complete.

Model: opus-5-5

internal/storage/conformance_test.go

@@ -14,8 +14,9 @@ import (
 
 // runStorerConformance is the shared Storer contract. Every backend that
 // can run in-process is expected to pass it: TestFileStorer runs it against
-// file://, TestS3Storer against s3://. A new backend inherits this coverage
-// by passing its own constructor, so the contract is defined once.
+// file://, TestS3Storer against s3://, TestRcloneStorer against rclone's
+// local backend. A new backend inherits this coverage by passing its own
+// constructor, so the contract is defined once.
 //
 // It exercises the public Storer interface: round-trip, stat, list with
 // prefix filtering, overwrite, delete, delete-of-missing, and not-found on