tiablob/tiasync: investigate supporting the publishing of smaller blobs

[misko9]

If Celestia is congested, we may need to publish less blobs or even smaller blobs. Validators will need to be able to detect this congestion. The first reduction could be to publish less blocks/blobs per pay for blob message. The next reduction could be to publish parts of the block, similar to how consensus distributes blocks in 64KB parts. Some research here is needed.