feat(metrics): add v1 metrics search and fix v2 query request body

[MichaelVessia]

What does this PR do?

Fixes the v2 metrics query request body (which was causing 400 Bad Request for
all queries) and adds a new metrics search subcommand backed by the v1
QueryMetrics API, following the same convention as logs search (v1) /
logs query (v2).

After this PR:

# v1 API, classic query syntax:
pup metrics search --query="avg:system.cpu.user{*}" --from="1h"

# v2 API, now works correctly:
pup metrics query --query="avg:system.cpu.user{*}" --from="1h"

Motivation

pup metrics query returns 400 Bad Request for every query. The v2
TimeseriesFormulaQueryRequest body was missing two required fields:

1. Name on the query: each MetricsTimeseriesQuery must have a name
   (e.g. "a") so formulas can reference it.
2. Formulas array: the request must include at least one formula
   referencing the named query (e.g. {Formula: "a"}).

Repro:

pup auth login
pup auth status  # confirm: authenticated, token valid

pup metrics query --query="avg:system.cpu.user{*}" --from="1h"
# => Error: failed to query metrics: 400 Bad Request (status: 400)

The existing code (cmd/metrics.go) was:

metricsQuery := datadogV2.NewMetricsTimeseriesQuery(datadogV2.METRICSDATASOURCE_METRICS, queryString)
timeseriesQuery := datadogV2.MetricsTimeseriesQueryAsTimeseriesQuery(metricsQuery)

This sets DataSource and Query but leaves Name nil and never populates
Formulas. The fix matches the official SDK example at
examples/v2/metrics/QueryTimeseriesData.go.

Additional Changes

Test improvements (cmd/metrics_test.go): Removed duplicate test case,
added wantErrContains assertions to verify error message content, replaced
brittle string-literal checks in TestMetricsSearchCmd with behavioral
assertions (subcommand registration, flag existence), added
TestMetricsCmd_Subcommands.

Additional Notes

Tested manually against a live Datadog account after building from this branch:

v1 search (new command):

$ pup metrics search --query="avg:system.cpu.user{*}" --from="1h"
{
  "from_date": 1770777733000,
  "query": "avg:system.cpu.user{*}",
  "res_type": "time_series",
  "resp_version": 1,
  "series": [
    {
      "aggr": "avg",
      "display_name": "system.cpu.user",
      "expression": "avg:system.cpu.user{*}",
      "interval": 20,
      "length": 180,
      "metric": "system.cpu.user",
      "pointlist": [[1770777740000, 12.03], [1770777760000, 10.95], ...]
    }
  ]
}

v2 query (fixed, previously 400'd):

$ pup metrics query --query="avg:system.cpu.user{*}" --from="1h"
{
  "data": {
    "attributes": {
      "series": [{"group_tags": [], "query_index": 0}],
      "times": [1770777740000, 1770777760000, ...],
      "values": [[12.03, 10.95, ...]]
    }
  }
}

Help output confirms both subcommands with API version labels:

$ pup metrics --help
Available Commands:
  query       Query time-series metrics data (v2 API)
  search      Search metrics (v1 API)

Unit tests: 13 test functions, 28 sub-tests, 0 failures.

Checklist

• The code change follows the project conventions (see CONTRIBUTING.md)
• Tests have been added/updated (if applicable)
• Documentation has been updated (if applicable)
• All CI checks pass
• Code coverage is maintained or improved

Related Issues

[MichaelVessia]

Hey 👋 , got a little carried away - not sure if v1 support is wanted but I saw it was there for some other APIs so I followed suit. Can adjust to decrease the scope if you want.