[FEATURE] Update dynamic scoring formula

[jdabtieu]

Currently we have max pts, min pts, num users before decay, but we should also have a variable controlling decay rate that can be adjusted based on how many participants are expected

The default (hard coded) is 11, but something like 40 works a lot better for a contest like BxMCTF (500 teams). This should be adjustable

Also, these parameters should be maybe editable mid-contest could cause issues with concurrency.

We should match CTFd Good news is that we can keep the three db columns in place. Bad news is that since CTFd uses the score_users function differently we might have to do some approximating or something for migration.

tldr:

Let that be f(solves). Then the point value becomes max(score_min, ceil(f(solves)))

Todo:

• rejudge function
• update dyn score function
• figure out migration

[jdabtieu]

diff for bxmctf:

diff --git a/src/helpers.py b/src/helpers.py
index 663ccc2..0448291 100644
--- a/src/helpers.py
+++ b/src/helpers.py
@@ -197,10 +197,10 @@ def update_dyn_score(contest_id, problem_id, update_curr_user=True):
         cid=contest_id, pid=problem_id))
     N_min = check[0]["score_min"]
     N_max = check[0]["score_max"]
-    N_users = check[0]["score_users"]
-    d = 11 * math.log(N_max - N_min) + N_users
-    old_points = min(math.ceil(math.e**((d - solves + 1) / 11) + N_min), N_max)
-    new_points = min(math.ceil(math.e**((d - solves) / 11) + N_min), N_max)
+    N_users = 20 # check[0]["score_users"]
+    d = 40 * math.log(N_max - N_min) + N_users
+    old_points = min(math.ceil(math.e**((d - solves + 1) / 40) + N_min), N_max)
+    new_points = min(math.ceil(math.e**((d - solves) / 40) + N_min), N_max)
     point_diff = new_points - old_points
 
     # Set new point value of problem