Recal: Fixing the Recal function bug

[KristinaGomoryova]

This PR deals with fixing the Recal function bugs and covers multiple issues.

• three new arguments were added to the processKnown function: type (A molecule type of the annotation), multiplier (The multiplier for the num column) and element_mass (The mass of the group used for correction), as well as the original type argument was renamed to type_col (The column name to store the type of compound).
• the type_col column is now computed using the multiplier (1 for O, 2 for H): step_result[[type_col]] <- step_result[[type_col]] + step_result[[num_col]]*multiplier
• paste function to create the form column was not working, so it was corrected as follows:

step_result <- dplyr::mutate(
    step_result,
    form=paste(C, H, O, N, S, P, E, S34, N15, D, Cl, Fl, Cl37, M, NH4, POE, NOE, sep="_")
  )

• columns were incorrectly deleted in here: step_result <- step_result[-remove_indices] . Now only `[-remove_indices] is used instead of original [-remove_indices, ]
• processKnown function is now used also if cols == 3, not only cols == 2

Accompanying test and test data to replicate the bug are also included.

Related issues:

• closes Recal: Error... replacement has 17 rows, data has 1050 #47
• closes MFAssignR: Recal function fails on MFAssignR model data in negative mode galaxytools#576