that if you pick one typical colony you are likely to pick a good one, but if you scrape a plate where some contaminant might be lurking, you might miss it in the crowd.

let say, you have transformed your bacteria with some plasmid. In such a case, the amplification and ligation did not have to work well and really each colony can have sligthly different version, with some mutations. If you pick only one colony, it probably comes from only one original cell and should be homogenous, but can differ from other colonies on the plate

Jack is true.
But from a frozen stock (or even one agar) if everything has been done correctly, in all likeliness all colonies are equivalent (there might be mutants, but that happen also within the colony, so it is not very relevant in general).
So you can use more than one colony if the experiment requires it. Just take multiple isolated colonies, to make sure that you are not picking something else at the same time. For example if you try to isolate E. coli from a soil containing some Proteus sp. multiple repeated steps of picking isolated colonies on fresh plate will be necessary to get rid of the contaminants.

Theoretically, every colony is clonal, stemming from single isolated cell that grew up to being visible macroscopically. Jack’s post are correct.