Support JSON response for /metrics endpoint, add client method for metrics

[arnavgarg1]

This PR adds a new metrics method to both the async and sync LoRAX client objects that returns a nearly equivalent payload to what the /metrics endpoint returns.

By default, the /metrics endpoint returns a response in plain/text format and in a format compatible with Prometheus. This PR modifies that endpoint to support a new optional query parameter called format that can optionally be set to JSON or just requires the application/json header.

curl -X GET "http://127.0.0.1:8080/metrics" -H "Accept: application/json"

On the Python client side, you can just call client.metrics(format="json") to get the relevant response back:

from lorax import Client

endpoint_url = "http://127.0.0.1:8080"
client = Client(endpoint_url)

metrics = client.metrics(format="json") # just call .metrics() if you want it in prometheus format

The structure follows this format:

{
        parent_metric_name: {
                "type": histogram | gauge | summary | counter
                "data": [
                     {"key": X, "value": Y},
                     {"key": X, "value": Y},
                     ...
                ]
        }
}

For metrics that are histograms, you will also find a sum and count key in the data objects.

Here's an example snippet when grabbing this output:

{'lorax_batch_current_max_tokens': {'data': [{'key': '', 'value': 0.0}],
                                    'type': 'gauge'},
 'lorax_batch_current_size': {'data': [{'key': '', 'value': 0.0}],
                              'type': 'gauge'},
 'lorax_batch_inference_count': {'data': [{'key': 'lorax_batch_inference_count{method="decode"}',
                                           'value': 508.0},
                                          {'key': 'lorax_batch_inference_count{method="prefill"}',
                                           'value': 4.0}],
                                 'type': 'counter'},
.....
 'lorax_request_skipped_tokens': {'data': [{'key': 'lorax_request_skipped_tokens{quantile="0"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="0.5"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="0.9"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="0.95"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="0.99"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="0.999"}',
                                            'value': 0.0},
                                           {'key': 'lorax_request_skipped_tokens{quantile="1"}',
                                            'value': 0.0},
                                           {'key': 'sum', 'value': 0.0},
                                           {'key': 'count', 'value': 512.0}],
                                                         'type': 'summary'}
....
}