feat: add prometheus metrics to STAC operations

Instrument register_stac.py and augment_stac_item.py with Prometheus
metrics for production observability.

Metrics:
- stac_registration_total: track create/update/skip/replace operations
- stac_http_request_duration_seconds: STAC API latency
- preview_generation_duration_seconds: augmentation timing
- preview_http_request_duration_seconds: preview API latency

SLOs: success >99%, STAC API <500ms, preview <10s

Docs: docs/prometheus-metrics.md with queries, alerts, dashboards

Author: wietzesuijker

docs/prometheus-metrics.md

@@ -0,0 +1,100 @@
+# Prometheus Metrics
+
+## Metrics Collected
+
+Pipeline scripts expose Prometheus metrics for observability. Metrics server runs on port 8000 in workflow pods.
+
+### STAC Registration (`register_stac.py`)
+```python
+stac_registration_total{collection, operation, status}
+# operation: create|update|skip|replace
+# status: success|error
+# Track failures, operation distribution
+
+stac_http_request_duration_seconds{operation, endpoint}
+# operation: get|put|post|delete
+# endpoint: item|items
+# STAC API latency, set SLOs
+```
+
+### Preview Generation (`augment_stac_item.py`)
+```python
+preview_generation_duration_seconds{collection}
+# Augmentation performance by collection
+
+preview_http_request_duration_seconds{operation, endpoint}
+# operation: get|put
+# STAC API response times during augmentation
+```
+
+## Key Queries
+
+**Success Rate (SLO: >99%)**
+```promql
+sum(rate(stac_registration_total{status="success"}[5m])) / sum(rate(stac_registration_total[5m]))
+```
+
+**Errors by Collection**
+```promql
+sum(rate(stac_registration_total{status="error"}[5m])) by (collection)
+```
+
+**STAC API Latency P95 (SLO: <500ms)**
+```promql
+histogram_quantile(0.95, rate(stac_http_request_duration_seconds_bucket[5m])) by (operation)
+```
+
+**Preview Duration P95 (SLO: <10s)**
+```promql
+histogram_quantile(0.95, rate(preview_generation_duration_seconds_bucket[5m])) by (collection)
+```
+
+**Throughput (items/min)**
+```promql
+sum(rate(stac_registration_total[5m])) * 60
+```
+
+## Setup
+
+Prometheus scrapes via PodMonitor (deployed in `platform-deploy/workspaces/devseed*/data-pipeline/`).
+
+**Verify:**
+```bash
+kubectl port-forward -n core svc/prometheus-operated 9090:9090
+# http://localhost:9090/targets → "geozarr-workflows"
+```
+
+## Grafana Dashboards
+
+- **Overview**: Success rate, throughput, error rate by collection
+- **Performance**: P95 latencies (STAC API, preview generation)
+- **Capacity**: Peak load, processing rate trends
+
+## Alerts
+
+**High Failure Rate**
+```yaml
+expr: rate(stac_registration_total{status="error"}[5m]) / rate(stac_registration_total[5m]) > 0.1
+for: 5m
+# Check STAC API status, verify auth tokens
+```
+
+**Slow Preview Generation**
+```yaml
+expr: histogram_quantile(0.95, rate(preview_generation_duration_seconds_bucket[5m])) > 60
+for: 10m
+# Check TiTiler API or asset access
+```
+
+**STAC API Latency**
+```yaml
+expr: histogram_quantile(0.95, rate(stac_http_request_duration_seconds_bucket[5m])) > 1
+for: 10m
+# Database overload or network issues
+```
+
+## SLOs
+
+- **Success Rate**: >99%
+- **STAC API P95**: <500ms
+- **Preview P95**: <10s