fix: optimize NFS mount options for Syncthing performance

- Replace defaults with explicit optimized NFSv3 options
- noatime: eliminates 8M+ atime setattr calls during Syncthing scans
- rsize/wsize=1048576: 32x larger transfer buffers (was 32KB)
- timeo=14: 1.4s timeout instead of 60s (LAN, not WAN)
- retrans=5: more retries since timeo is low

Root cause: NFS server only supports NFSv3, and 'defaults' uses
kernel-minimum rsize/wsize=32KB + atime updates on every read.
Syncthing's 80K-file scans generated 52M getattr + 8M setattr RPCs.
These options reduce metadata RPC overhead and increase data throughput.

modules/nixos/filesystem/hoardingcow-mount.nix

@@ -7,7 +7,16 @@
     fileSystems."/mnt/HoardingCow_docker_data" = {
       device = "192.168.1.2:/WorkHorse_docker_data";
       fsType = "nfs";
-      options = [ "defaults" "nofail" "_netdev" ];
+      options = [
+        "noatime"
+        "hard"
+        "nofail"
+        "_netdev"
+        "rsize=1048576"
+        "wsize=1048576"
+        "timeo=14"
+        "retrans=5"
+      ];
     };
   };
 }