Rationalize samples

[craigbox]

$ rg bookinfo --glob '*.md' --files-with-matches | wc -l
      60
$ rg sleep --glob '*.md' --files-with-matches | wc -l
      61
$ rg notsleep --glob '*.md' --files-with-matches | wc -l
       2
$ rg httpbin --glob '*.md' --files-with-matches | wc -l
      40
$ rg helloworld --glob '*.md' --files-with-matches | wc -l
      35

(The fact our curl client container is called sleep has always perplexed me.)

I would like the docs to be able use one sample deployment throughout. Could we please have an (optional) add-on for Bookinfo that includes a pod called "client" and a pod called "server", where client is curl and server is httpbin, if that's still the gold standard?

(httpbin is unmaintained - consider https://github.com/mccutchen/go-httpbin / https://hub.docker.com/r/mccutchen/go-httpbin)

[craigbox]

I discussed this with @linsun, @dhawton and @peterj this morning, and we think this would be a good path forward:

• rename "sleep" to "curl" — don't use 'client' and 'server' (Add curl sample to replace sleep istio#53381)
• have a bookinfo-ambient.yaml or bookinfo-all.yaml file which includes bookinfo, bookinfo-versions, curl and httpbin.

[craigbox]

#15767