[data] Fix early stop for multiple limit ops.

[raulchen]

Why are these changes needed?

• When there are multiple limit ops, the dataset should stop when the smallest limit is reached. For example, ds.limit(1000).limit(1) should stop when 1 row is outputted. But the current implementation won't. This PR fixes this issue.
• Also mark LimitOp as throttling_disabled, so it will be prioritized in the scheduler.

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• 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 added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• 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 :(

[scottjlee]

LGTM, by the way, i previously implemented this limit fusion logic for the case of Limit[n] -> Limit[m] to Limit[min(n, m)]:

ray/python/ray/data/_internal/logical/rules/limit_pushdown.py

Line 97 in 98768c2

def _apply_limit_fusion(self, op: LogicalOperator) -> LogicalOperator:

But this is currently disabled since LimitPushdownRule is not used right now. We could also enable this logic separately to further optimize this case