dlt/faultline/compare

Compare occurrences of the same group to spot which inputs vary between successes and failures.

1. Parse $ARGUMENTS: first token is the group id, optional second token is how many occurrences to compare (default 3, max 10).
2. Call recent_occurrences with group_id: <id> and limit: <count> to get a list of occurrence ids.
3. For each id, call get_occurrence to pull the full payload (backtrace, locals, request params, headers).
4. Diff across the occurrences and surface what changes:

• request_params — which keys are present, which values differ
• local_variables — same: which keys, which values
• request_url / request_method — different endpoints hitting the same fingerprint?
• user_identifier — same user repeatedly, or many distinct users?
• backtrace — should be near-identical; flag if a frame differs (means the fingerprint may be over-grouping).

1. Render a compact comparison: per varying field, show the value from each occurrence.
2. Lead with the highest-signal difference. Captured locals usually beat request params for diagnosing what triggers the bug.

If only one occurrence exists for the group, say so — there's nothing to compare. Filtered values will appear as [FILTERED] and shouldn't be flagged as differences (Rails.application.config.filter_parameters + config.sanitize_fields).