[Docker] Start building ray-ml CPU Docker image again

[amogkam]

Why are these changes needed?

Related issue number

Checks

• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[krfricke]

Looks good, and build seems to succeed. Thanks!

[amogkam]

FYI @javi-redondo @alanwguo @richardliaw

[ijrsvt]

@amogkam this looks good

Looks like ray-ml:nightly is still the same as ray-ml:nightly-gpu (this is the intended behavior right)!!

build-docker-images.py --py-versions py36 --device-types cpu cu101 cu102 --build-type BUILDKITE --build-base: 

These tags will be created for ray-ml:  defaultdict(<class 'list'>, {'nightly-py36-cpu': ['nightly-py36-cpu', 'ec7760-py36-cpu']})

and

build-docker-images.py --py-versions py36 --device-types cu110 cu111 cu112 --build-type BUILDKITE --build-base:

These tags will be created for ray-ml:  defaultdict(<class 'list'>, {'nightly-py36-cu112': ['nightly-py36-cu112', 'nightly-py36-gpu', 'nightly-py36', 'nightly-py36-gpu', 'ec7760-py36-cu112', 'ec7760-py36-gpu', 'ec7760-py36', 'ec7760-py36-gpu']})

Also holds for py37 & py38!

[amogkam]

@ijrsvt yes this was the intention!