Hi Paul! This is not my field of research so take my remarks cautiously but from my understanding in prokaryotes the initiator formyl methionyl-tRNA is always used at the start of synthesis, whatever the start codon is. Recognition of the start codons is dependent on the Initiator Factors, especially IF3 which ensures that the formyl methionyl-tRNA is paired to correct start codons. Thus, all proteins should be starting with M. But in fact, that is quite irrelevant since the initial Methionine is often removed in the final, mature peptide. Should IF3 amounts decrease, initiation can start at non conventional initiation codons (this is the regulatory mechanism that modulates the amounts of IF3 in the cell). In some cases, non-formylated methionyl-tRNA can be used instead of the formyl methionyl-tRNA to start translation, but this comes with a cost to the bacterium (this is a resistance mechanisms to actinonin antibiotic).

Regarding which codons can be used as start codons in eukaryotes, it seems that the diversity of start codons could be greater than what is usually assumed (see for example https://academic.oup.com/nar/article/45/7/3615/2990259). It could be interesting to see whether translation always starts with a M in all cases, whether the initiator formyl methionyl-tRNA is always used, and what IF does initiation at those start codons depends on. Protein sequencing to verify which amino-acid is present would be tempting, except that in most cases that amino-acid will be removed in the final peptide and thus will not be present in the sequence.