perf(webapp): index EnvironmentVariableValue.environmentId

The /api/v1/projects/:projectRef/envvars/:slug/:name loader calls
`getEnvironmentWithRedactedSecrets`, which runs a Prisma findMany on
EnvironmentVariableValue filtered by environmentId + isSecret. The
only existing indexes are the primary key and a unique on
(variableId, environmentId), so environmentId is never the leading
column. Postgres plans a Parallel Seq Scan over the whole table
(roughly 2.2M rows / 400MB heap) to find an average of 24 matching
rows.

Adding a btree on environmentId alone flips the plan to an index scan
(~24 row average per environment, 536 max in our sample), which removes
the CPU + buffer-pin (LWLock:BufferContent) contention this query
generates on the writer.

The composite (variableId, environmentId) unique stays where it is;
this is purely additive.

Author: ericallam

.server-changes/env-var-value-environment-id-index.md

@@ -0,0 +1,6 @@
+---
+area: webapp
+type: improvement
+---
+
+Add an index on `EnvironmentVariableValue(environmentId)` so the env-var lookups behind `GET /api/v1/projects/:projectRef/envvars/:slug/:name` use an index scan instead of a parallel sequential scan over the whole table.

internal-packages/database/prisma/migrations/20260520120000_add_environment_variable_value_environment_id_idx/migration.sql

@@ -0,0 +1,3 @@
+-- CreateIndex
+CREATE INDEX CONCURRENTLY IF NOT EXISTS "EnvironmentVariableValue_environmentId_idx"
+  ON "EnvironmentVariableValue"("environmentId");

internal-packages/database/prisma/schema.prisma

@@ -2020,6 +2020,7 @@ model EnvironmentVariableValue {
   lastUpdatedBy Json?
 
   @@unique([variableId, environmentId])
+  @@index([environmentId])
 }
 
 model Checkpoint {