[Core] Fix check failure due to negative available resource

[jjyao]

Why are these changes needed?

The sequence of events that can trigger the check failure:

1. A node has PG resources {"CPU_group_aaa": 2, "CPU_group_0_aaa": 1, "CPU_group_1_aaa": 1}
2. Worker1 acquires 1 PG CPU resource, node available resource becomes {"CPU_group_aaa": 1, "CPU_group_0_aaa": 0, "CPU_group_1_aaa": 1}
3. Worker1 calls ray.get() which temporarily released its CPU resource, node available resource becomes {"CPU_group_aaa": 2, "CPU_group_0_aaa": 1, "CPU_group_1_aaa": 1}
4. Worker2 acquires 1 PG CPU resource, node available resource becomes {"CPU_group_aaa": 1, "CPU_group_0_aaa": 0, "CPU_group_1_aaa": 1}
5. ray.get() returns from worker1 and it acquires back the resource, node available resource becomes {"CPU_group_aaa": 0, "CPU_group_0_aaa": -1, "CPU_group_1_aaa": 1}
6. Worker3 acquires 1 PG CPU resource and the check failure happens since it can get CPU_group_1_aaa but not CPU_group_aaa. The fix is that we allow CPU_group_aaa to go negative as well to match the indexed PG CPU resources.

Related issue number

Closes #50433

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 :(

[MengjinYan]

Thanks!