Should be easier to write tests against it

Tests run in e2e mode will more closely match the normal node
operation. Notably: full randomx and normal poa1 vs. poa2 difficulty.

There are still a lot of constants that are reset in this mode -
*mostly* to speed up the test run and allow it to work on smaller
blockchains. But if material differences are discovered we can
reset some of those constants back to more representative values
as needed

…ess"

This reverts commit d209fb2.

Co-authored-by: Lev Berman <[email protected]>

This commit adds the packing difficulty in `v2_index_data_size_by_packing`
metric.

see: ArweaveTeam/arweave-dev#568

This gives a truer estimate of the miner's expected read performance

This fixes an issue where read performance was severely degradeed
when multiple storage modules were stored on a single disk. This
was particularly bad for SATA, although was noticeable on SAS disks
as well.

Only current (active) peers should be able to receive the
chunks, others will generate unnecessary timeouts. In fact
using the active peers from the last 24 hours can also be
a great solution if we assume a peer seen during the last
24 hours will probably be come back quick.

When chunks are propagated across the network, the to be
acknowledged are the trusted peer. On the previous version,
the trusted peer where put at the end of the list.

Cleaned up ar_tx_emitter_sup, converting old tuple to map.

Refacto and added comments and documentation.

see: ArweaveTeam/arweave-dev#720

When a server has the full transaction locally, and uses
spinning disk, the amount of requests from other nodes is
too high, and leads to a bottleneck on rocksdb database.

This commit adds a semaphore to control the amount of
requests on `ar_http_iface_middleware:handle_get_tx/3` function,
to avoid a node to be overflooded by requests.

For the first ~2 months after the 2.8 HF activation nodes
may have less than the required 3 months of reward history.
This is okay.

…tected`

…tected`