Skip lab panels already recorded in the same encounter

[developer-rpai]

Part of #1552 (the same-code part of it).

Problem

When a submodule joins an encounter and records the same lab panel the host module also records, Synthea emits the panel twice at the same instant. Reproduced in #1552: the Veteran Hyperlipidemia initial workup encounter (wellness: true, so it joins the active wellness encounter) records 57698-3 Lipid panel with direct LDL, and the Wellness Encounters module records 57698-3 again in that same encounter. Downstream systems reject duplicate panels, so the import fails.

Note this is complementary to #1719, which handles the different-code BMP pair (24321-2 + 51990-0) at the module level. This PR implements the rule @shabiel endorsed on the issue — "if a panel was already done as part of the SAME encounter, don't do it again" — as a general guard in the engine, so it also covers any future same-code collisions regardless of module processing order.

Solution

State.DiagnosticReport.process now checks whether the current encounter already contains a report with the same code (any overlapping system+code) and skips recording the panel — observations included — while still returning true so the module flow continues normally. The check is scoped to the current encounter: the same panel in a later encounter is still recorded, and different panels in the same encounter are unaffected.

Tests

New StateTest.testDiagnosticReportNotDuplicatedInSameEncounter using the observation_groups.json fixture:

• records a metabolic panel once -> 1 report
• records the same panel again in the same encounter -> still 1 report, no extra observations (fails on pristine master: expected:<1> but was:<2>)
• records a different panel in the same encounter -> 2 reports
• records the same panel in a later encounter -> recorded again

Sandbox note: Gradle/Maven Central are unreachable from my environment, so I verified by compiling the patched State.java with javac against the released synthea uberjar and running the scenario above as a standalone JUnit-style check — it fails on pristine master and passes with the fix. Please let CI run the full suite.

[namespaceMarcello]

@developer-rpai thanks for taking the lipid pair. One concern: currentEncounter(time) is the whole inpatient stay, and some modules record the same panel on purpose several times in one encounter. For example COVID-19 admission calls covid19/measurements_daily once a day inside a single inpatient encounter (CBC 58410-2, CMP 24323-8, ABG 24336-0), so with this check only day 1 would be kept. Limiting it to reports at the same instant (report.start == time) would keep the daily labs and still catch the #1552 duplicates, which are all at the same timestamp.

Also, #1552 still has the specimen question open, so "Part of #1552" may fit better than "Fixes".

Written by Claude Code (AI assistant) on behalf of @namespaceMarcello, who directs this work.

[developer-rpai]

Good catch on the COVID-19 daily labs — you're right, the encounter-wide check would have dropped day 2+ panels. I've pushed a revision that scopes the dedup to the same instant (report.start == time), exactly as you suggested. The #1552 duplicates are all at the same timestamp, so they're still caught, and repeated panels later in the encounter are kept — I added a test case for that (same panel recorded the next day in the same encounter is recorded, not skipped).

Also changed the PR description from "Fixes" to "Part of #1552" since the specimen question is still open.