Overhaul Key API about ink HOT 8 CLOSED

This commit adds std::ops::{Add, AddAssign} implementations for Key for u32 and u64.

• Signed versions will follow.
• Concerns about adding another std::ops::Sub and std::ops::SubAssign implementation.
• Also we could add something similar to a ptrdiff with std::ops::Sub<Key> for Key impl. However, what return type should it have?

from ink.

Commits 8880390 and 10e5f15 remove Key::with_chunk_offset and Key::with_offset in favor of the new core::ops::{Add, AddAssign} implementations.

from ink.

Commit 42c490b marks Env::load as unsafe fn since it compares itself with a pointer deref in Rust which is also known as being an unsafe operation.

Changes to Env::store and Env::clear will follow with the same reasoning.

from ink.

Commit bb2e8ee marks the other Env::{store, clear} fns as unsafe as the previous commit.

This shall guide users of pdsl_core to use the safer cell, chunk and collections abstractions.

from ink.

Commit 89b0be5 adds core::ops::Sub impl for Key and the KeyDiff type.
Look at commit 4509914 on how to use this new interface.

Currently missing are some unit tests for KeyDiff.

Edit: Commit b91fb4a adds simple unit tests for KeyDiff.

from ink.

Missing pieces:

• impl core::ops::Sub<{u32, u64, u128}> for Key
• impl core::ops::Add<u128> for Key
• impl core::ops::Add<{i32, i64, i128}> for Key. Note that this hasn't been made yet since the unsigned variants can be implemented a bit more efficiently. This might not be transparent to users so could lead to a slight performance degradation when using signed instead of unsigned after implementing this. Besides that these trait implementations are for convenience. This requires a byte_utils::sign_extend procedure to sign extend the i32, i64 or i128 to the 32 byte of a key in order to subtract or add from or to it.

from ink.

Besides other things this commit adds core::ops::{Sub, SubAssign} implementations for Key.

Also during implementation we found out that core::ops::{Add,AddAssign,Sub,SubAssign} implementations for RHS = {i32,i64,i128} do not make sense and are semantically difficult to articulate on Key instances so it is better to leave them away and stay with the unsigned variants that have a semantically clear usage story.

So after removing this work item what is missing is implementations of core::ops::{Add,AddAssign,Sub,SubAssign} ofu128 for Key and also a conversion impl from KeyDiff to u128.

from ink.

The remaining parts have been implemented in 966d29d, bc711cb and 09b7bbd.

from ink.