[SOL] Fallback to default implementation of shouldReduceLoadWidth

[LucasSte]

Problem

When I was working on the PR for eBPF upstream, I noticed the shouldReduceLoadWidth function that is there to appease the kernel verifier. If we remove it, we have some gains.

Benchmarks

Programs/sbf

  SBF program                          expected actual  diff
  solana_sbf_rust_128bit                    630   630    +0 ( +0%)
  solana_sbf_rust_alloc                    4767  4767    +0 ( +0%)
  solana_sbf_rust_custom_heap               324   324    +0 ( +0%)
  solana_sbf_rust_dep_crate                  20    20    +0 ( +0%)
  solana_sbf_rust_iter                     1513  1513    +0 ( +0%)
  solana_sbf_rust_many_args                1281  1281    +0 ( +0%)
  solana_sbf_rust_mem                      1195  1194    -1 ( -0%)
  solana_sbf_rust_membuiltins                69    68    -1 ( -1%)
  solana_sbf_rust_noop                      342   342    +0 ( +0%)
  solana_sbf_rust_param_passing             108   108    +0 ( +0%)
  solana_sbf_rust_rand                      301   301    +0 ( +0%)
  solana_sbf_rust_sanity                  13576 13576    +0 ( +0%)
  solana_sbf_rust_secp256k1_recover       88376 88376    +0 ( +0%)
  solana_sbf_rust_sha                     21984 21984    +0 ( +0%)

No changes to Solana program or Pinocchio entrypoint.

p-token instructions:

Instruction	Master	This PR	Diff
initialize_mint	108	108	0
initialize_account	157	157	0
initialize_multisig	221	220	-1
transfer	76	76	0
approve	136	135	-1
revoke	102	101	-1
set_authority	130	129	-1
mint_to	122	122	0
burn	129	129	0
close_account	120	120	0
freeze_account	150	149	-1
thaw_account	146	145	-1
transfer_checked	105	105	0
approve_checked	183	182	-1
mint_to_checked	176	175	-1
burn_checked	147	147	0
initialize_account2	179	178	-1
initialize_account3	247	247	0
initialize_multisig2	342	341	-1
initialize_mint2	229	229	0
amount_to_ui_amount	424	423	-1
ui_amount_to_amount	697	696	-1
initialize_immutable_owner	38	38	0
sync_native	62	62	0
batch	1493	1493	0