We want to ensure that there are no major discrepancies between the statuses from l2 and the statuses from the emc-check runs.  This plot shows the correlation between the statuses, for all towers and all fills (in which status/ped tables are generated from both l2 and the emc-check runs). 

We mostly care about the cases in which a good status is assigned in l2 and a bad status in emc-check, and vice versa.  The following two 1D histograms show this more clearly.  (Only cases in which the status in l2 and emc-check disagree are shown, otherwise there would be a huge spike at 1 in both plots.)

Status in l2 when status in emc-check = 1
status     counts
0              831
2              270
18            450
36            39
38            93
54            12
72            84
90            9
110          1
136          70
138          1
154          21
200          2

Status in emc-check when l2 status = 1
status      counts
2              187
4              4
18            688
36            6
38            1
54            1
72            2
136          6
154          2
254          323

We need to check each of these cases to see which code (l2status.py for l2 and emc-check) is assigning the 'right' status. 

l2 status = 0, emc-check status = 1
This is a problem we already knew about, in that the emc-check code doesn't know whether the tower is masked out at l2.  Once this problem is fixed, it may also solve most of the issue with the hot (status = 2) towers.

l2 status = 1, emc-check status = 254