Instruction log parsing could be better sanitized

[sintemal]

Currently, the explorer parses the hierarchy of inner instructions based on the program logs:
https://github.com/solana-labs/explorer/blob/7b6116dfdda770c82fa41190cb52b0ffba8f329e/app/utils/program-logs.ts#L91C49-L91C49

Especially the parsing of program returns based on log.includes("success") and log.includes("failed") seems a bit flawed,
as the logs of native programs do not have the Program log: prefix like normal programs and, therefore, do not get filtered out beforehand.
This means that if a native program emits a log message containing success or failure, the parsing is incorrect.
One could achieve this by using
https://github.com/solana-labs/solana/blob/0cd57addcf4ae242e0d068d09fc0359a4249a5e6/programs/zk-token-proof/src/lib.rs#L33

or, -as the zk-token-proof program doesn't seem to be deployed yet- by grinding a Pubkey containing the string and using one of the various other logs: https://github.com/solana-labs/solana/blob/2210af60ee615eda44bd3afc0b2093c23600a005/program-runtime/src/invoke_context.rs#L600

I guess the best fix here would be using a regex similar to the parsing of an invoked program.