feat(client): make Kubernetes API client QPS and burst configurable

[oneubauer]

Why is this PR required? What issue does it fix?:

Follow-up to #321 (configurable worker threads). #321 made the provisioner's
worker count tunable via OPENEBS_IO_WORKER_THREADS, but all workers share a
single Kubernetes clientset, and that clientset's rate limiter is still pinned
to the client-go defaults (QPS 5 / burst 10). Raising the worker count alone
therefore just creates more goroutines that contend for the same client-side
throttle — under bursty PVC create/delete load the provisioner self-throttles
before the API server's own API Priority & Fairness ever comes into play.

This PR exposes the client's QPS and burst so operators running at scale can
raise them alongside workers.

What this PR does?:

Adds two new, optional environment variables (and matching Helm values) to tune
the rate limiter on the shared Kubernetes API client:

• OPENEBS_IO_CLIENT_QPS: maximum sustained queries/sec to the API server.
• OPENEBS_IO_CLIENT_BURST: maximum burst above QPS.

Wiring:

• cmd/provisioner-localpv/app/env.go: env-var names + getClientQPS() /
  getClientBurst() getters (functional menv lookup, same pattern as the
  existing worker-thread/timeout getters).
• pkg/kubernetes/client/client.go: WithQPS() / WithBurst() functional
  options applied to the rest.Config before the clientset is built.
• cmd/provisioner-localpv/app/start.go: reads the env vars and passes the
  options through when constructing the client.
• Helm chart: localpv.controller.clientQPS / localpv.controller.clientBurst
  values, conditionally injected into the Deployment and DaemonSet env, plus
  README parameter docs.

Default behavior is unchanged. When the variables are unset (or empty), no
override is applied and client-go's own defaults (QPS 5 / burst 10) remain in
effect. The env vars are only injected by the chart when explicitly set, so
existing deployments render identically.

Does this PR require any upgrade changes?:

No. The feature is opt-in and defaults preserve prior behavior.

If the changes in this PR are manually verified, list down the scenarios covered::

• Unset vars → helm template renders no OPENEBS_IO_CLIENT_QPS /
  OPENEBS_IO_CLIENT_BURST env entries; client uses client-go defaults.
• Vars set via localpv.controller.clientQPS / clientBurst → both env vars
  rendered into the pod spec (verified for both Deployment and DaemonSet/node
  modes), and the provisioner logs the overridden QPS/burst at startup.
• Invalid / non-positive values → ignored, falls back to client-go defaults.

Any additional information for your reviewer?:

Continuation of #321 — that PR made the worker pool configurable; this PR makes
the shared client's rate limiter configurable so the added workers aren't
bottlenecked by the default client-side throttle. QPS is a float32 and burst
an int to match the rest.Config.QPS / rest.Config.Burst field types in
client-go.

Checklist:

• Fixes #
• PR Title follows the convention of <type>(<scope>): <subject>
• Has the change log section been updated?
• Commit has unit tests
• Commit has integration tests
• (Optional) Are upgrade changes included in this PR? If not, mention the issue/PR to track:
• (Optional) If documentation changes are required, which issue on https://github.com/openebs/openebs-docs is used to track them:

[niladrih]

nit:

Suggested change

	func getClientQPS() (float32, bool) {
	val, present := menv.Lookup(ProvisionerClientQPS)
	if !present || val == "" {
	return 0, false
	}
	qps, err := strconv.ParseFloat(val, 32)
	if err != nil || qps <= 0 {
	return 0, false
	}
	return float32(qps), true
	}
	func getClientQPS() (float32, bool) {
	val, present := k8sEnv.GetFloat64(ProvisionerClientQPS, 0.0)
	if !present || val <= 0.0 {
	return 0.0, false
	}
	return float32(val), true
	}

[niladrih]

nit:

Suggested change

	val, present := menv.Lookup(ProvisionerClientBurst)
	if !present || val == "" {
	return 0, false
	}
	burst, err := strconv.Atoi(val)
	if err != nil || burst <= 0 {
	return 0, false
	}
	return burst, true
	val, present := k8sEnv.GetInt(ProvisionerClientBurst, 0)
	if !present || val <= 0 {
	return 0, false
	}
	return val, true

[niladrih]

nit:

Suggested change

	var clientOpts []mKube.OptionFn
	clientOpts := make([]mKube.OptionFn, 0, 2)

[niladrih]

Not required, as we're already doing this in applyRateLimits

Suggested change

	if qps > 0 {
	c.qps = qps
	}
	}
	}
	// WithBurst sets the maximum burst of queries above the QPS limit the client
	// may send to the Kubernetes API server. A non-positive value is ignored,
	// leaving the client-go default in place.
	func WithBurst(burst int) OptionFn {
	return func(c *Client) {
	if burst > 0 {
	c.burst = burst
	}
	c.qps = qps
	}
	}
	// WithBurst sets the maximum burst of queries above the QPS limit the client
	// may send to the Kubernetes API server. A non-positive value is ignored,
	// leaving the client-go default in place.
	func WithBurst(burst int) OptionFn {
	return func(c *Client) {
	c.burst = burst

[niladrih]

Suggested change

	clientQPS: ""
	# Maximum burst of queries above clientQPS the provisioner allows to the Kubernetes API
	# server. Leave empty to use the client-go default (10).
	clientBurst: ""
	client:
	qps: ""
	# Maximum burst of queries above clientQPS the provisioner allows to the Kubernetes API
	# server. Leave empty to use the client-go default (10).
	burst: ""