[Feature] Pull-cid mode optimization - don't re-import the same deals

[jcace]

• Currently pull-cid mode will simply scan the filesystem and attempt to request every deal that matches the car files
• This is inefficient as some deals may have already been attempted before, or may be already sealed
• We should pre-fill a cache list of deals attempted/in-progress/sealed per dataset, and ignore those when running pull-cid mode