Timeout consolidation

[shawkins]

There's a recognition on several issues that there are too many timeouts that are not all well defined.

On eclipse-jkube/jkube#2024 it was agreed that writeTimeout should be removed as primary concern.

The other timeouts:

1. Config/RequestConfig requestTimeout - applied as the HttpClient readTimeout and enforced at OperationSupport.waitForResult
2. Config/RequestConfig scaleTimeout - used for scale operations when the wait boolean is specified. Wth the rolling logic deprecated Config / RequestConfig rollingTimeout is unused #4826 the internal call to scaling will be removed, so the only place will be from the user scale call.
3. Config/RequestConfig websocketTimeout - the readTimeout specifically for websocket requests - used two different ways. Watches set it as the readTimeout on the client, while exec waits on the initial future for that amount of time.
4. readyWaitTimeout / logWaitTimeout - many pod operations check if the pod is in a good state prior to performing the operation
5. delete timeout - uses a common context timeout value, turns the delete into a blocking operation - applied for each delete when operating over a collection
6. openshift buildconfig timeout - also uses the common context timeout value - applies to how long it takes to upload the build and get the response back
7. Config/RequestConfig connectionTimeout - applied as the connection timeout

Let's discuss how to collapse / combine these.