about merging multiallelics

[vicbp1]

Hi,
Thank you for developing this amazing tool.
I understand that splitting multiallelic is needed before phasing. However, I observed a weird pattern after phasing: I have unphased variants with two alternative alleles.

Unphased
chr21 5292305 G A 1/0
chr21 5292305 G T 0/1

Phased

chr21 5292305 . G A 1|0
chr21 5292305 . G T 1|0

I observed this for several multiallelics, and when I collapse into one variant, this genotype becomes unphased because of the incompatibility of phases: 1/2

How do you think I should avoid this problem?

Thanks

Vic