chore(analysis/calculus/mean_value): use `𝓝[Ici x] x` instead of `𝓝[Ioi x] x` (#5472)
In many parts of the library we prefer `𝓝[Ici x] x` to `𝓝[Ioi x]
x` (e.g., in assumptions line `continuous_within_at`). Fix MVT and
Gronwall's inequality to use it if possible.
Motivated by #4945