feat:aggregate_diff value sub

[clickme-zxy]

when write aggregate diff, (https://numpy.org/doc/stable/reference/generated/numpy.diff.html) the aggregate diff will sub the two value and panic "attmpt to panic sub overflow"

the value we has is :i8、i6、i32 、i64 when come to i64， the aggregate will panic of overflow(since the largest value type is i64)

so implement the value sub so to avoid the panic

[MichaelScofield]

Arithmetic overflow is a global issue all around our codes. We should find a global way to handle it adequately.

[MichaelScofield]

Two possible ways I can think of right now are:

1. Expanding our Value to support u128 and i128 primitive types, and make sure our column schema supports max to u64 and i64; then change the Primitive::LargestType of u64 and i64 to u128 and i128.
2. Handling overflow error globally, which requires a global error handler somewhere.

[tisonkun]

@MichaelScofield @clickme-zxy, could you provide a reproduction so that we can understand the issue more intuitively?

Said a SQL that encounters error while it should not be.

[MichaelScofield]

@tisonkun create another issue https://github.com/GreptimeTeam/greptimedb/issues/3385 for this.

[MichaelScofield]

Close as superceded by #3385