fix: Input batch size overflow vulnerability

[yinggeh]

What does the PR do?

1. Cast uint32_t (batch_size_) to int32_t is not safe. Cast to int64_t instead. Before the change, request with overflowed batch size will bypass core check and propagate to backend. Although backend will catch the error, it's still better to stop the request at early stage.
2. Remove input_memory_type from ValidateBytesInputs parameters as it's not used.

Checklist

• PR title reflects the change and is of format <commit_type>: <Title>
• Changes are described in the pull request.
• Related issues are referenced.
• Populated github labels field
• Added test plan and verified test passes.
• Verified that the PR passes existing CI.
• Verified copyright is correct on all changed files.
• Added succinct git squash message before merging ref.
• All template sections are filled out.
• Optional: Additional screenshots for behavior/output changes with before/after.

Commit Type:

Check the conventional commit type
box here and add the label to the github PR.

• fix

Related PRs:

triton-inference-server/server#8165

Where should the reviewer start?

Test plan:

L0_input_validation--base

• CI Pipeline ID:
  27387023

Caveats:

Background

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

• closes GitHub issue: #xxx

[yinggeh]

batch_size_ is uint32_t type.

[yinggeh]

No HTTP or gRPC client test because input size is gigantic. The minimum input byte size to validate, assuming bool input with shape [-1,1], is 2^31 bytes or 2 GB.

In gRPC client, the maximum byte size allowed is exactly 2^31-1 bytes.
In http client, request stops at server with error
tritonclient.utils.InferenceServerException: [400] inference header size should be in range (0, -2147483525), got: 155.
In shared memory, returns error when shared memory size is 1 GB.

Bus error (core dumped)
UserWarning: resource_tracker: There appear to be 1 leaked shared_memory objects to clean up at shutdown

[ziqif-nv]

LGTM