Set the SF16 pure NNUE reference nps wrt SF11

Fishtest with Stockfish 11 had 1.6Mnps as reference nps
and 0.7Mnps as threshold for the slow worker.
Set the new reference nps to 0.64Mnps according to the 60% slowdown
of the SF 16 wrt SF 11 on the ARCH=x86-64-bmi2
Set the new threshold for slow worker to 0.231Mnps according to
the 67% slowdown of the SF 16 wrt SF 11 on the ARCH=x86-64-sse41-popcnt

compiler clang++ 16.0.6

- arch=bmi2 (Intel Xeon CPU E5-2680 v3)
```
sf_base =  1515099 +/-   6276 (95%)
sf_test =   610755 +/-   2686 (95%)
diff    =  -904343 +/-   5940 (95%)
speedup = -59.689% +/- 0.392% (95%)
```

- arch=sse41-popcnt (Intel core i7 3770k)
```
sf_base =  1929595 +/-  19733 (95%)
sf_test =   654161 +/-   5421 (95%)
diff    = -1275433 +/-  14999 (95%)
speedup = -66.099% +/- 0.777% (95%)
```

Raise worker version to 207, also server side.

server/fishtest/api.py

@@ -25,7 +25,7 @@
 on how frequently the main instance flushes its run cache.
 """
 
-WORKER_VERSION = 206
+WORKER_VERSION = 207
 
 
 def validate_request(request):

server/fishtest/rundb.py

@@ -523,7 +523,7 @@ def aggregate_unfinished_runs(self, username=None):
                     nps += concurrency * task["worker_info"]["nps"]
                     if task["worker_info"]["nps"] != 0:
                         games_per_minute += (
-                            (task["worker_info"]["nps"] / 1184000)
+                            (task["worker_info"]["nps"] / 640000)
                             * (60.0 / estimate_game_duration(run["args"]["tc"]))
                             * (
                                 int(task["worker_info"]["concurrency"])

server/utils/delta_update_users.py

@@ -47,7 +47,7 @@ def compute_games_rates(rundb, info_tuple):
     # 1184000 nps is the reference core, also set in rundb.py and games.py
     for machine in rundb.get_machines():
         games_per_hour = (
-            (machine["nps"] / 1184000)
+            (machine["nps"] / 640000)
             * (3600.0 / estimate_game_duration(machine["run"]["args"]["tc"]))
             * (int(machine["concurrency"]) // machine["run"]["args"].get("threads", 1))
         )

worker/games.py

@@ -1290,15 +1290,15 @@ def parse_options(s):
         games_concurrency * threads,
     )
 
-    if base_nps < 462000 / (1 + math.tanh((worker_concurrency - 1) / 8)):
+    if base_nps < 231000 / (1 + math.tanh((worker_concurrency - 1) / 8)):
         raise FatalException(
             "This machine is too slow ({} nps / thread) to run fishtest effectively - sorry!".format(
                 base_nps
             )
         )
     # 1184000 nps is the reference core benched with respect to SF 11,
     # also set in rundb.py and delta_update_users.py
-    factor = 1184000 / base_nps
+    factor = 640000 / base_nps
 
     # Adjust CPU scaling.
     _, tc_limit_ltc = adjust_tc("60+0.6", factor)

worker/sri.txt

@@ -1 +1 @@
-{"__version": 206, "updater.py": "PHFUVXcoxBFiW2hTqFN5q5WdAw2pK8uzFC7hyMUC3cLY5bGZPhL8TBGThtqDmcXd", "worker.py": "R8rJ4c9ZQTbuSu80ayVyMs7D6/TOVR5qMn6R04XCsBp4BTkCKXfy7hSrpMaLa0nX", "games.py": "D+5sFwptbwAhvn+VM4oHyRwqUPfGUWLQPFbWHA/uK77rTUsU/r49gBMjjz6k5Ncq"}
+{"__version": 207, "updater.py": "PHFUVXcoxBFiW2hTqFN5q5WdAw2pK8uzFC7hyMUC3cLY5bGZPhL8TBGThtqDmcXd", "worker.py": "6v8r4DGKbGAS8joczJrNjO3xX4/QImrKkU4xOpNXc5PGy4kFNgcyvxGc96FsUVww", "games.py": "LI8f72sm9yoqxO0sJPbpXTSGIgVfJhb0WuKPZ2AU2z0QCz/4IYnfYik+B/NiWG7x"}

worker/worker.py

@@ -54,7 +54,7 @@
 # Several packages are called "expression".
 # So we make sure to use the locally installed one.
 
-WORKER_VERSION = 206
+WORKER_VERSION = 207
 FILE_LIST = ["updater.py", "worker.py", "games.py"]
 HTTP_TIMEOUT = 30.0
 INITIAL_RETRY_TIME = 15.0