implementing mempool and remove 1 tx per tick limitation

[Franziska-Mueller]

Currently, qubic protocol has these limitation about transactions & mempool:

• One address (ID) can only have 1 transaction(tx) per tick. Eg: if you submit 2 transactions to tick N: A sends QUs to B and A sends QUs to C, only one of them will be included (picked by tick leader).
• One address (ID) can only have 1 transaction(tx) in the mempool. Eg: if you submit 2 transactions update #1 A sends QUs to B at tick N and Update efi-build-develop.yml #2 A sends QUs to C at tick (N+1). Tx Update efi-build-develop.yml #2 will replace update #1 because it has higher scheduled tick.
• Recently we allow computor ID to have 128 txs per tick (they have their own mempool):
  

We now want to remove above limitation and allow all IDs to have many txs per tick. (thus many txs in the mempool)

Structure of Transaction (in network_messages/transaction.h):

struct Transaction
{
    m256i sourcePublicKey;
    m256i destinationPublicKey;
    long long amount;
    unsigned int tick; // <~ scheduled tick
    unsigned short inputType;
    unsigned short inputSize;
}

A mempool should be able to do these 3 basic functions (just suggestion, you can improve while doing this):

void update(Transaction* tx, enum strategy):

• check valid and add tx to the pool. It maybe unnecessary to add a tx that has scheduled_tick = current_tick + 2000
• To avoid spamming to the pool, we need a strategy (to add/remove which tx) to keep total number of tx per tick less than or equal NUMBER_OF_TRANSACTIONS_PER_TICK (1024 atm)
• For now strategy is just random:
  

(future tx picking strategy)

Transaction* get(unsigned int tick, int index):

• get a transaction from the pool of tick (already sorted by strategy above), return nullptr if there is no more txs

void prune(unsigned int tick):

• Delete all info before tick
• reorganize to save mem space

[J0ET0M]

waiting for #246