Use same type for block number

[PropzSaladaz]

Description

• Right now we treat block numbers as different integer types in different parts of the code, so there isn’t one clear “source of truth” for what range is actually valid
• In practice, the system can only go as high as the most restrictive type in the path, which puts the current ceiling at roughly 4 billion blocks (about 4.29B).
• That mismatch is risky at high block heights: values can be narrowed or collide with special marker values, which can lead to incorrect behavior across components.

Some concrete examples of block-number types currently used:

• unsigned (via BlockNumber)
• int (used in some parsing/temporary paths)
• uint64_t (used in storage/progress-related code)
• int64_t (used in some VM/execution context fields)
• size_t (used in a few progress/callback signatures)
• u256 (appears in error-info/metadata contexts)