On Devnet-12 after blobs spamming DB size is significantly bigger than for other ELs

[kamilchodola]

As per EF DevOps:

Noticed that some of our machines (arms) are running out of disk (devnet 12). Those machines come with very little free disk to begin with, so its very easy to notice. However, looks like there is a pattern about nodes that run nethermind run out of disk a lot quicker than those that don't run nethermind.

nethermind pairs with free disk space:
teku: 35GB free
nimbus: 31GB free
lodestar: 8GB free
lighthouse: 4.5GB free

Seems like Geth and Besu are taking only around 20GB of disk space where Nethermind takes from 70-100GB

Looking at detailed db size breakdown:

45G blobTransactions
35M blockInfos
34M blockNumbers
3.4G blocks
126M bloom
12M canonicalHashTrie
19M code
1.9M discoveryNodes
124M headers
20M metadata
8.0K peers
1.4G receipts
53G state
12M witness