Split df into multiple dfs by row limit, keeping entries of some matching value together in same df?

Question

Simplified df:

    group  to_ignore
0       5        100
1       3        103
2       3        104
3       2        115
4       3        126
5       4