fix: reduce costs on kinesis streams

[vasco-santos]

We are currently being billed quite a bit in AWS for all the Kinesis streams we keep open for dev/CI.

This PR addresses this by:

• not deploying 2 kinesis streams per env to keep historical content. We really just need the production stream still deployed until we read it.
• we can also decrease the persistence of the streams to keep costs down

As follow ups, we should also consider (I will create issues):

• tool that proactively tears down running infrastructure for stalled PRs (or setup tests in a way that kinesis is put on and off for the tests)
• get historical content kinesis stream flushed out

[vasco-santos]

Minimum value supported

[seed-deploy]

View stack outputs

• pr231-w3infra-CarparkStack

  Name	Value
  BucketName	carpark-pr231-0
  Region	us-east-2

• pr231-w3infra-RoundaboutStack

  Name	Value
  ApiEndpoint	https://xgp69ho7x3.execute-api.us-east-2.amazonaws.com
  CustomDomain	https://pr231.roundabout.web3.storage

• pr231-w3infra-SatnavStack

  Name	Value
  BucketName	satnav-pr231-0
  Region	us-east-2

• pr231-w3infra-UploadApiStack

  Name	Value
  ApiEndpoint	https://6n39mbjl8l.execute-api.us-east-2.amazonaws.com
  CustomDomain	https://pr231.up.web3.storage

• pr231-w3infra-BusStack

• pr231-w3infra-ReplicatorStack

• pr231-w3infra-UcanInvocationStack

• pr231-w3infra-UploadDbStack

[travis]

this is great, thanks @vasco-santos!