Stateless HTTP Indexing sources

Is your feature request related to a problem? Please describe.
We want to write data via the http endpoint and we have a very large amount of data and need to scale out indexing nodes.
The current Ingesting API is pretty good, but there are a few issues: this is a stateful endpoint, which requires queues to be maintained in local disk. when the http request returns a successful response, the data is not actually written to the index, but to the queue. If the node/disk hangs, the data is lost, but the client doesn't know that.

Describe the solution you'd like
Provide a stateless HTTP endpoint. which don't have local storage, once the handler received Indexing Request, handler just call the indexing flow and waiting for the result, then send the result to http client.
HTTP indexing source don't need to maintain the checkpoint, just indexing all the data in the request and then return the result.

Client--> Quickwit HTTP endpoint -> Handler--> indexing actor-> upload split-> send response to http client

Also provide the a write buffer to improve the write throughput

Client->Quickwit HTTP endpoint -> Handler -> Write Buffer(Set the threshold based on time and size) -> indexing actor -> upload split -> send response to http client.

Describe alternatives you've considered

Another option would be to replica the queues in current ingesting API, but this would be more complicated.

Start by reading the existing Ingesting API, indexing flow, and queue/checkpoint behavior. Trace how an HTTP request reaches the indexing actor and how results are returned. Done means the issue's stateless endpoint and optional size- or time-based write buffer are specified and validated without relying on local queue storage.