Send documents by batch instead of one by one

[brunoocasali]

Description
Basically, instead of pushing documents one by one, let's send them as a batch so Meilisearch can process them in batches, increasing the performance and the overall indexation time.

[curquiza]

⚠️ I want to add a warning here: we don't want to push too big batches to avoid memory crash issues by sending too big batches. I think 10 documents at the same time would be a good start. I would decrease by 10 the task queue without impacting too much the RAM, even if the documents are big.

A second iteration could be to add an option in this library to set up the size of the batch.