#2459 Update Service Levels docs to match the current UI

Summary

Some statements in the Service Levels guide no longer match the Logfire UI. This PR corrects them:

• Reliability history span. The guide said that the history defaults to the last 24 hours and that you choose a span only on the Burn rate view. Both views start with the full rolling window of the target, and the span choice (24 hours, 7 days when the window is longer than 7 days, or the full window) applies to both.
• Investigating failures. The guide referred to an Investigate failing events button that opens Explore. The detail page now has Open in SQL Workbench and, for record-based targets, Open in Live view. Both use the range that you select by dragging across the chart, or the displayed range when there is no selection. SQL Workbench opens the failing records for a record-based target, or the metric rows for a metric-based target. When a burn-rate alert has fired, its row has an Investigate link that opens the chart around the most recent firing.
• Histogram metrics. The guide said that a histogram target counts values below a threshold. You can choose which side of the threshold is good (Below is good or Above is good).
• Infrastructure as Code. The released Terraform and Pulumi providers have an experimental logfire_slo resource. The Infrastructure as Code resource list and the Service Levels guide now mention it. The docs name the resource only and do not describe its attributes.

The changes do not touch the burn-rate tier section, which #2458 updates separately. The two PRs merge cleanly in either order.

Preview

I built both pages locally with the unified docs preview (make docs-serve flow), together with #2458. Both pages render, and every link on them resolves: the new links to Infrastructure as Code, SQL Workbench and Live view, the in-page anchors, and the Terraform, OpenTofu and Pulumi registry links. Please still check the hosted preview.