Your point only applies to those with a recombinant vector. But even on those, Oxford were so fast because they already had a developed platform and they were working on a very similar epitope.
Spending a bit more time would have been surely safer, and would not have delayed things significantly. What we have now might not be good enough after all, if it's true that the SA variant escapes immune recognition on individuals vaccined with Oxford-AZ. I hope to be wrong.