core/txpool/blobpool: upgrade blobpool

[mask-pp]

Due to the large size of blob transactions, the goal of this PR is to reduce unnecessary repeated operations of extracting, storing, and retrieving blob transactions. As a result, only the transaction metadata is stored in limbo.
In the event of a reorganization, transactions will not be deleted from blobpool.store. Instead, their metadata will be transferred to limbo storage. When the finalization occurs, blobpool.store will be triggered to delete these transactions.

[cskiraly]

@mask-pp , was there a specific reason you made everything in blobTxMeta public?

[mask-pp]

@mask-pp , was there a specific reason you made everything in blobTxMeta public?

This PR aims to avoid deleting from BlobPool.store and then storing to limbo.store when reorg appears. If the reorg appears, the tx content still stored in BlobPool.store and just needs to change the relate metadata in limbo.
But currently, I found this PR can can't work due to billy.Database cell is static, if not this PR can works well.

[mask-pp]

@mask-pp , was there a specific reason you made everything in blobTxMeta public?

This PR aims to avoid deleting from BlobPool.store and then storing to limbo.store when reorg appears. If the reorg appears, the tx content still stored in BlobPool.store and just needs to change the relate metadata in limbo. But currently, I found this PR can can't work due to billy.Database cell is static, if not this PR can works well.

Maybe we change the size of limbo.store, but it's a break change. I'm not sure whether it's worth doing or not.

[cskiraly]

The data flow in the original code is not ideal for larger throughput, I agree.

My question is however specific to your change of making everything public in blobTxMeta, which is like 90% of the lines changed by the PR. Is that necessary for some reason?

[mask-pp]

The data flow in the original code is not ideal for larger throughput, I agree.

My question is however specific to your change of making everything public in blobTxMeta, which is like 90% of the lines changed by the PR. Is that necessary for some reason?

Oh sorry, this PR is not mature. change blobTxMeta wants to reuse the struct so that the object can be encoded and stored in the database.
I need to know how billy.Database works at first, and then update this PR.

[mask-pp]

Hi friends @cskiraly and @rjl493456442, sorry to bother you, the PR is combined into avoid_restore_blob_tx. Plz review again, tks.