Add BBolt rate limiting and tune Badger defaults for large archives (v0.48.12)

- Increase Badger cache defaults: block 512→1024MB, index 256→512MB
- Increase serial cache defaults: pubkeys 100k→250k, event IDs 500k→1M
- Change ZSTD default from level 1 (fast) to level 3 (balanced)
- Add memory-only rate limiter for BBolt backend
- Add BBolt to database backend docs with scaling recommendations
- Document migration between Badger and BBolt backends

Files modified:
- app/config/config.go: Tuned defaults for large-scale deployments
- main.go: Add BBolt rate limiter support
- pkg/ratelimit/factory.go: Add NewMemoryOnlyLimiter factory
- pkg/ratelimit/memory_monitor.go: New memory-only load monitor
- CLAUDE.md: Add BBolt docs and scaling guide

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

app/config/config.go

@@ -41,9 +41,9 @@ type C struct {
 	EnableShutdown      bool          `env:"ORLY_ENABLE_SHUTDOWN" default:"false" usage:"if true, expose /shutdown on the health port to gracefully stop the process (for profiling)"`
 	LogLevel            string        `env:"ORLY_LOG_LEVEL" default:"info" usage:"relay log level: fatal error warn info debug trace"`
 	DBLogLevel          string        `env:"ORLY_DB_LOG_LEVEL" default:"info" usage:"database log level: fatal error warn info debug trace"`
-	DBBlockCacheMB      int           `env:"ORLY_DB_BLOCK_CACHE_MB" default:"512" usage:"Badger block cache size in MB (higher improves read hit ratio)"`
-	DBIndexCacheMB      int           `env:"ORLY_DB_INDEX_CACHE_MB" default:"256" usage:"Badger index cache size in MB (improves index lookup performance)"`
-	DBZSTDLevel         int           `env:"ORLY_DB_ZSTD_LEVEL" default:"1" usage:"Badger ZSTD compression level (1=fast/500MB/s, 3=default, 9=best ratio, 0=disable)"`
+	DBBlockCacheMB      int           `env:"ORLY_DB_BLOCK_CACHE_MB" default:"1024" usage:"Badger block cache size in MB (higher improves read hit ratio, increase for large archives)"`
+	DBIndexCacheMB      int           `env:"ORLY_DB_INDEX_CACHE_MB" default:"512" usage:"Badger index cache size in MB (improves index lookup performance, increase for large archives)"`
+	DBZSTDLevel         int           `env:"ORLY_DB_ZSTD_LEVEL" default:"3" usage:"Badger ZSTD compression level (1=fast/500MB/s, 3=balanced, 9=best ratio/slower, 0=disable)"`
 	LogToStdout         bool          `env:"ORLY_LOG_TO_STDOUT" default:"false" usage:"log to stdout instead of stderr"`
 	LogBufferSize       int           `env:"ORLY_LOG_BUFFER_SIZE" default:"10000" usage:"number of log entries to keep in memory for web UI viewing (0 disables)"`
 	Pprof               string        `env:"ORLY_PPROF" usage:"enable pprof in modes: cpu,memory,allocation,heap,block,goroutine,threadcreate,mutex"`
@@ -124,9 +124,9 @@ type C struct {
 	Neo4jMaxTxRetrySeconds int `env:"ORLY_NEO4J_MAX_TX_RETRY_SEC" default:"30" usage:"max seconds for retryable transaction attempts"`
 	Neo4jQueryResultLimit  int `env:"ORLY_NEO4J_QUERY_RESULT_LIMIT" default:"10000" usage:"max results returned per query (prevents unbounded memory usage, 0=unlimited)"`
 
-	// Advanced database tuning
-	SerialCachePubkeys  int `env:"ORLY_SERIAL_CACHE_PUBKEYS" default:"100000" usage:"max pubkeys to cache for compact event storage (default: 100000, ~3.2MB memory)"`
-	SerialCacheEventIds int `env:"ORLY_SERIAL_CACHE_EVENT_IDS" default:"500000" usage:"max event IDs to cache for compact event storage (default: 500000, ~16MB memory)"`
+	// Advanced database tuning (increase for large archives to reduce cache misses)
+	SerialCachePubkeys  int `env:"ORLY_SERIAL_CACHE_PUBKEYS" default:"250000" usage:"max pubkeys to cache for compact event storage (~8MB memory, increase for large archives)"`
+	SerialCacheEventIds int `env:"ORLY_SERIAL_CACHE_EVENT_IDS" default:"1000000" usage:"max event IDs to cache for compact event storage (~32MB memory, increase for large archives)"`
 
 	// Connection concurrency control
 	MaxHandlersPerConnection int `env:"ORLY_MAX_HANDLERS_PER_CONN" default:"100" usage:"max concurrent message handlers per WebSocket connection (limits goroutine growth under load)"`