The ability to flashing the firmware is an attack vector, as is the existence of firmware in the first place.
Flashing firmware is not an attack vector, if the firmware is signed by a trusted party. An ASIC would be far worse; there are no ways for third parties to truly audit the ASIC. Open source code, on the other hand, can be audited by a large swath of programmers.
with specific attention payed to avoiding information leaks that could lead to a successful side channel attack.
What side channel attacks are you specifically worried about on a hardware wallet? The only opportunity for a hardware wallet to leak secret information through a side channel is during signing. When the hardware wallet is secured by a password, this will only occur when the user is using the device for signing legitimate transactions. This will occur infrequently, to put it lightly. Power and timing side channel attacks require huge sample sizes, relative to the number of transactions a user will ever have signed. Not to mention that it's impossible for the host to time the actual signing, unless it's correlating with power consumption and I don't believe a PC can extract that information from their USB host chips.