Paginate/compress VERY large workflows so they can travel through gRPC

## Summary

A user is experiencing an issue when querying for a VERY large workflow (135k+ nodes) using the API.

```
curl http://localhost:2747/api/v1/workflow-events/pre-production-testing-1?listOptions.fieldSelector=metadata.name=otpgal-gqpcf
{"error":{"grpc_code":8,"http_code":429,"message":"trying to send message larger than max (296376209 vs. 104857600)","http_status":"Too Many Requests"}}
```

This message limit is placed by gRPC. We should find a way to compress/paginate or otherwise modify the message so it's able to travel over gRPC. At the very least we should provide a reasonable error to this problem.

Note that this only affects sending workflows over the API. CLI and execution are fully unaffected.

## Diagnostics

What Kubernetes provider are you using?

All platforms

What version of Argo Workflows are you running?

All versions

---

**Message from the maintainers**:

Impacted by this bug? Give it a 👍. We prioritise the issues with the most 👍.

Start by reproducing the API request for a workflow with 135k+ nodes and tracing where the gRPC maximum-message error occurs. Done means large workflows can travel through the API via compression or pagination, or the API returns a reasonable error when they cannot; CLI and execution must remain unaffected.

Written by the indexing model from the issue text.