improv : optimize ix compute units and cleanup

[Nagaprasadvr]

1. use derive_pda_address when we are just validating pdas
2. remove redundant checks
3. remove accounts which arent required and few checks
4. Add cu tracker
5. show cus and cu diffs using a python script
6. remove example

[zfedoran]

This looks awesome, give me some time to review it

[zfedoran]

Looking through this, one thing I'd like to keep is the pda calculations in dev/test. I think what I have here is a good compromise. Take a look at this comment for the why.

What are your thoughts on this?

Regarding the simple hash over the on-curve point method, I do really like that and would prefer to use it over the more expensive is_on_curve checked version that find_program_address ships with.

[Nagaprasadvr]

Cool i will add those tests back

ig we can just move thse pda find fncs inside the cfg tests module wyt ? so they are scoped only inside tests #[cfg(debug_assertions)] pub fn archive_pda() -> (Pubkey, u8) { Pubkey::find_program_address(&[ARCHIVE], &crate::id()) }

[Nagaprasadvr]

Looking through this, one thing I'd like to keep is the pda calculations in dev/test. I think what I have here is a good compromise. Take a look at this comment for the why.

What are your thoughts on this?

Regarding the simple hash over the on-curve point method, I do really like that and would prefer to use it over the more expensive is_on_curve checked version that find_program_address ships with.

Yeah actually we don't have to check if it's on curve while validating but only while creating the account so we have the canonical bump

[Nagaprasadvr]

@zfedoran let me know what do u think, more cus to optimize we can do it parallely as we develop. But steel is so cus hungry