fix: use default maxBytes and ncclBuffSize for p4 running all-to-all

[ytsssun]

Issue #, if available:
#690

Description of changes:
Use default maxBytes and ncclBuffSize when running all-to-all test case on P4 instances.

Test done
Ran the all-to-all test case on P4 with this change. The test passed:

    mpi_test.go:123: Multi node job completed
--- PASS: TestMPIJobPytorchTraining (40.41s)
    --- PASS: TestMPIJobPytorchTraining/multi-node:alltoall_perf (40.41s)
        --- PASS: TestMPIJobPytorchTraining/multi-node:alltoall_perf/MPIJob_succeeds (40.31s)
PASS
ok      github.com/aws/aws-k8s-tester/test/cases/nvidia 52.345s

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[ytsssun]

Addressed the comment