fix(getter): redundancy getter cleanup

[istae]

Checklist

• I have read the coding guide.
• My change requires a documentation update, and I have done it.
• I have added tests to cover my changes.
• I have filled out the description and linked the related issues.

Description

1. removed close call
2. remove ctx in the struct antipattern
3. inflight requests are canceled when recovery is initialized.
4. the joiner does not construct a redundancy getter when fallback mode is false and the strategy is none.

Open API Spec Version Changes (if applicable)

Motivation and Context (Optional)

Related Issue (Optional)

Screenshots (if appropriate):

[zelig]

we should fix this test not skip

[zelig]

what if neither of these conditions are met. this loop then will never terminate... it is possible m can be less than shard+parity and also the fetches started via Get are concurrent and can reach the counts later then checked here

[istae]

every time c returns, it's guaranteed that either fetchedCnt or failedCnt is already incremented (inside fetch) so eventually one of the conditions are met.

if m is less than shard + parity, then a previous run of fetches has already finished and incremented the counters.

[zelig]

think about this once more. Because of the single flight pattern and Get calls interleaving prefetch fetches, there is a decent probability that the loop will not terminate. All you need to imagine is that the shardCnt-th successful fetch is.
I think we should trigger closing the initRecovery channel when we increment the count (in one atomic call)

[zelig]

there is no need to run this routine for every fetcgh.
creating context and cancel hould be done onky once

[istae]

context is canceled is a separate goroutine when the recovery is initialized...this is how we achieve early termination.
it's safe to call cancel twice in separate goroutines.

[zelig]

it does not make sense to wait on 128 parallel routines to do what you can do in one. Create context with cancel in the prefetch fetch function which is cancelled only once. And apply your logic only to the fetch calls initiqted within Get calls.

[NoahMaizels]

Last week:

• Published guide to node state and health endoints and reorganized troubleshooting section https://docs.ethswarm.org/docs/bee/working-with-bee/troubleshooting
• Published new content in erasure coding section including cost calculator: https://docs.ethswarm.org/docs/develop/access-the-swarm/erasure-coding/
• Submitted backlog of issues to bee repo related to api problems / improvements
• Various 2.0 related comms (tweets, discord announcements)
• Working with Tamas on subtitles and editing for health and status api endpoints video guide

This week:

• Present new docs content and highlight video guides at SCC
• Adapt and add doubling and neighborhoods guide to docs
• Add devops guide to docs
• 2.0 comms
• Docs issues backlog catchup time permitting