Not mentioning, that i didn't expect this from you, I guess, if you would design a $5 toy, you wouldn't take a flash micro, would you?

One should always keep in mind, that there's a huge difference between an eprom based 87C51 and a flash based 89C51. It's very often reported in erratas, application notes and some datasheets (here not so often...), that data corruption took place in code memory, when brown-out events occured!!! So, it's not only a question of 'turn-off and turn-on the micro again', because power-on reset of poor RC circuit didn't properly initialize the micro. It's a question of code memory corruption! And here the manufacturer should much clearer focuse on, that an unproper reset circuit, which is not able to reset the micro during brown-out events, should definitely be AVOIDED in any case.
It was also reported in these publishings, that even the lock bits can become unlocked by brown-out events, if micro wasn't resetted properly!! That would be a desaster for a company needing to hide their code...

So, whenever flash micros are used a proper brown-out reset is a must!

Kai