how useful flat profile with mid-range DSLR shooting ?

I have this relatively generic question about the real benefit of shooting with a flat color profile on a mid-range DSLR with 8 bit 420 H264 encoding.
I understand the idea behind shooting e.g. with the flat cineform profile on a Canon EOS 90D is that it is supposed to provide a somewhat wider tonal range and hence more flexible color correction.
But since this flat video is recorded in 8 bit (in my case), more severe correction of tonal range and curves means more risk to create artefacts like banding, noise etc., as the narrow 8 bit range is designed to be "just enough" to encode a finalized clip.
camera sensor -> flat profile -> 8 bit encoding/recording -> tonal correction in post expands the [255] available discrete levels in some ranges to a point where we may get noticeable artefacts (and the sophisticated 32bit FP processing inside Resolve can't compensate for the limited amount of levels in the source material)
while:
camera sensor -> in-camera tonal correction -> 8 bit encoding/recording -> less correction needed in post, hence less risk of excessively stretching discrete levels
Is this reasoning correct ?
If so, I see a perhaps negative trade-off between better tonal range control and risk of more noise when shooting with a flat profile and 8 bit encoding.
I understand the idea behind shooting e.g. with the flat cineform profile on a Canon EOS 90D is that it is supposed to provide a somewhat wider tonal range and hence more flexible color correction.
But since this flat video is recorded in 8 bit (in my case), more severe correction of tonal range and curves means more risk to create artefacts like banding, noise etc., as the narrow 8 bit range is designed to be "just enough" to encode a finalized clip.
camera sensor -> flat profile -> 8 bit encoding/recording -> tonal correction in post expands the [255] available discrete levels in some ranges to a point where we may get noticeable artefacts (and the sophisticated 32bit FP processing inside Resolve can't compensate for the limited amount of levels in the source material)
while:
camera sensor -> in-camera tonal correction -> 8 bit encoding/recording -> less correction needed in post, hence less risk of excessively stretching discrete levels
Is this reasoning correct ?
If so, I see a perhaps negative trade-off between better tonal range control and risk of more noise when shooting with a flat profile and 8 bit encoding.