Explore implementing alarm thresholds

## Description

Create “more intelligent” alarms with a higher thresholds like multiple forms or multiple instances to help us detect if it’s a problem submitting forms vs. a specific path.

## Context

## Incident 1

[2026-03-17-no-checksums-on-files](https://docs.google.com/document/d/147IAprDWt7xEQdxPYs6BDQiiwNfPh6fIyuwICQC10Vo/edit?tab=t.0)

`Could not submit response for form {ID}. Received error: Unexpected Submission lambda processing result: {"status":false}`

`File references found but no checksums provided -- Severity level: 1`

`Content MD5 checksum is required to generate signed URL. $_key: form_attachments/2026-03-17/formID/attachmentID/filename.pdf`

### Incident 2

[2026-02-02-client-ip-detection-causing-hcaptcha-errors](https://docs.google.com/document/d/1fVSzj3jlPJmd9hcdIQYt1aePXlS6zP1WCp-_32R_hNc/edit?tab=t.0)

> When # submission attempts are blocked in X timeframe, alarm notifies team that something might be wrong.

In this case, an extra space was added to IPs on certain services blocking them from being able to submit their forms.

## Impact

This would help finetune alarms and ensure that SEV1 alarms don't go off at too low a threshold, such as a client testing form upload with saving progress.