fix: reduce memory_limit to 256M and configure PHP-FPM process management settings

docker/php/php.ini

@@ -1,5 +1,7 @@
 ; Production PHP settings for ClinicPro on Coolify
-memory_limit = 1024M
+; 256M is ample for API requests (uploads are capped at 32M below); the old 1024M
+; let a single leaky/looping request eat a quarter of a small VPS before dying.
+memory_limit = 256M
 upload_max_filesize = 16M
 post_max_size = 32M
 max_execution_time = 60

docker/php/zz-pool.conf

@@ -5,3 +5,14 @@
 ; wrong environment. Keep the container env so compose-provided vars reach PHP.
 [www]
 clear_env = no
+
+; Process manager: bounded pool + child recycling. Without pm.max_requests the
+; default children live forever and slow PHP memory leaks accumulate for hours
+; until the host OOMs (the messenger workers already recycle hourly; this gives
+; FPM the same hygiene).
+pm = dynamic
+pm.max_children = 8
+pm.start_servers = 2
+pm.min_spare_servers = 2
+pm.max_spare_servers = 4
+pm.max_requests = 500