Remove sender stakes from banking_stage buffer prioritization

[tao-stones]

Problem

It seems unnecessary to prioritize packets by sender stake on top of transaction's compute-unit price, due to:

1. Quic stake weight part already giving priority to highly staked senders;
2. When compute-unit price is widely adapted, due to micro-lamport as its unit, it is very unlikely enough transactions have same priority; makes adding sender stake into MinMaxHeap ordering unnecessary;

therefore it makes sense to remove sender stake from MinMaxHeap ordering to save some computing and simplify code a bit.

Summary of Changes

Remove sender-stake from MinMaxHeap ordering.

Fixes #

[nikhayes]

Agreed. You can see evidence that the few senders using fee priority are already using highly nuanced fees. There is incentive not to have the same priority as others and minutely tweak things... thus there will likely be a great variety of fee priorities observed.

Pays 0.000006001 Sol
https://explorer.solana.com/tx/3E5sjRJYV9E6vccWdjzRkxuYrZoTQ26NjcJhMkY8G3UWsEXY6eTad8sjebbRSUQ59TFc6vZHYrBAE4QT8VvBmQjh

Pays 0.000005002
https://explorer.solana.com/tx/2M4F8nKdmV8SqsWXKtEPWXhJxfbtACvMEbeGA45mLvWJ3fWTkPafZQQTRkyq2AGZLBS5uNj5bifTk3r3YNf3L17T

In summary, I don't think sorting by stake weight will accomplish much other than causing worry about stake centralization issues (i.e. the powerful get richer -- which I've seen people wonder already) and adding a fud angle.

Pull request has been modified.