minor stuff related to rate limit

.env

@@ -50,7 +50,6 @@ TALKNET_ROOT_CODE_DIRECTORY=/home/bt/dev/storyteller/storyteller-web/models/tts
 TALKNET_MODEL_CHECK_SCRIPT_NAME="vocodes_model_check_talknet.py"
 
 TTS_INFERENCE_SIDECAR_HOSTNAME=localhost:8000
-SIDECAR_MAX_SYNTHESIZER_MODELS=2
 
 #TTS_WAVEGLOW_VOCODER_MODEL_FILENAME=waveglow_256channels_universal_v5.pt # THIS SOUNDS HORRID
 #TTS_WAVEGLOW_VOCODER_MODEL_FILENAME=waveglow_256channels_v4_uberduck.pt # THIS IS TOO SMALL! 300MB vs 1.2 GB!

crates/service/storyteller_web/src/tts_inference_job.rs

@@ -776,6 +776,10 @@ async fn process_job(
       .virtual_model_lfu
       .insert_returning_replaced(tts_synthesizer_fs_path.to_str().unwrap_or(""))?;
 
+  if let Some(model_path) = maybe_unload_model_path.as_deref() {
+    warn!("Remove model from LFU cache: {:?}", model_path);
+  }
+
   let output_audio_fs_path = temp_dir.path().join("output.wav");
   let output_metadata_fs_path = temp_dir.path().join("metadata.json");
   let output_spectrogram_fs_path = temp_dir.path().join("spectrogram.json");