Not identifing H_abstraction family

Describe the bug
I tried to calculate the rate of the H_abstraction reaction : H2N2_116 + ammonia_5 <=> NH2_3 + H3N2_4, without including TS guess. I got this reaction from RMG output that identified it as H_abstraction. While running ARC the reaction wasn't identified as H_abstraction as expected.

How to reproduce
Here is some more examples that ARC didn't identified to be H_abstraction but they are.

job_types:
  rotors: false

allow_nonisomorphic_2d: True  

compute_thermo: false
freq_scale_factor: 1

species:
  - label: 'H4N2_1'
    smiles: 'NN'
  - label: 'H2_2'
    smiles: '[H][H]'
  - label: 'NH2_3'
    smiles: '[NH2]'
  - label: 'H3N2_4'
    smiles: '[NH]N'
  - label: 'ammonia_5'
    smiles: 'N'  
  - label: 'HN2_6'
    smiles: '[N]=N' 
  - label: 'N2H2_19'
    smiles: 'N=N'
  - label: 'H_8'
    smiles: '[H]'    
  - label: 'H2N2_116'
    smiles: '[NH][NH]'
  - label: 'H2N2_113'
    smiles: '[N]N'

reactions:
  - label: H2N2_116 + ammonia_5 <=> NH2_3 + H3N2_4
  - label: H2N2_116 + H4N2_1 <=> H3N2_4 + H3N2_4
  - label: N2H2_19 + H2N2_116 <=> HN2_6 + H3N2_4
  - label: H_8 + H3N2_4 <=> H2_2 + H2N2_116

ARC.log output:

Setting multiplicity of reaction NH2_3 + H3N2_4 <=> H2N2_116 + ammonia_5 to 3
Considering reaction: H2N2_116 + ammonia_5 <=> NH2_3 + H3N2_4



Setting multiplicity of reaction H3N2_4 + H3N2_4 <=> H2N2_116 + H4N2_1 to 3
Considering reaction: H2N2_116 + H4N2_1 <=> H3N2_4 + H3N2_4



Setting multiplicity of reaction HN2_6 + H3N2_4 <=> N2H2_19 + H2N2_116 to 3
Considering reaction: N2H2_19 + H2N2_116 <=> HN2_6 + H3N2_4



Setting multiplicity of reaction H2_2 + H2N2_116 <=> H_8 + H3N2_4 to 3
Considering reaction: H_8 + H3N2_4 <=> H2_2 + H2N2_116

Here is a graphic image of the species of the first reaction: