ESP32 Forum

rare instances of lags shortly after boot up

May be the I-cache still 'warming up'.
And/or other initialization activities (WiFi connect,...) running concurrently and hogging up the the flash's bandwidth.

You could enable the UART's loop back to send 'yourself' some dummy bytes first to have as much of the relevant code as possible loaded from flash before going to town.

Is it possible to receive a notification when the TX FIFO becomes empty?

There is a pending feature request about this.