Performance issue in well log consistency rule check for big arrays (20k + columns)

A function used by the mechanism to check consistency on the number of column appears not linear regarding total number of column to proceed.

Total number of column = number of column in the whole welllog dataframe. 1 column could be either a 1D curve or one column of a curve array, if the welllog contains 5 1D curve and one curve array with 7 columns, the total number of column in the bulk dataframe = 5x1 + 1x7 = 12 columns in total.

To treat 10k columns, the algorithm takes ~1,5s, 20k => 6s, 40k => 20s, 60k => 45s ...

Accoustic welllog can contains 100k+ columns.