Improve function_abi_no_longer_unwind performance by reordering current, baseline

[jw013]

Partially addresses #767

Comparing 0.29.1 to 0.30.0, it seems likely the largest performance issue is due to a new lint introduced in 0.30.0, function_abi_no_longer_unwind, which took 5 hours on my machine to check the aws ec2 crate using the command from the issue report. I ran every c-s-c release from 0.25.0 to 0.28.0 as well and did not see any other performance issues.

     Parsing aws-sdk-ec2 v1.41.0 (current)
      Parsed [ 128.600s] (current)
     Parsing aws-sdk-ec2 v1.41.0 (baseline)
      Parsed [  87.512s] (baseline)
    Checking aws-sdk-ec2 v1.41.0 -> v1.41.0 (assume minor change)
    Starting 61 checks, 6 unnecessary on 4 threads
...
        PASS [17804.320s]       major        function_abi_no_longer_unwind
...
     Checked [17812.879s] 61 checks; 61 passed, 6 unnecessary
    Finished [18041.203s] aws-sdk-ec2

I am not very knowledgeable about ABI's but if I understood things correctly from a cursory investigation, the default "Rust" ABI (i.e. the most common kind) is considered unwind-able. After reordering the current and baseline branches of the query to look for non-unwind-able functions first, the function_abi_no_longer_unwind lint runs in under a second:

     Parsing aws-sdk-ec2 v1.41.0 (current)
             Features: behavior-version-latest,default,rt-tokio,rustls,test-util
      Parsed [ 144.442s] (current)
     Parsing aws-sdk-ec2 v1.41.0 (baseline)
             Features: behavior-version-latest,default,rt-tokio,rustls,test-util
      Parsed [ 102.323s] (baseline)
    Checking aws-sdk-ec2 v1.41.0 -> v1.41.0 (assume minor change)
    Starting 67 checks, 6 unnecessary on 4 threads
...
        PASS [   0.605s]       major        function_abi_no_longer_unwind
...
     Checked [  21.817s] 67 checks; 67 passed, 6 unnecessary
    Finished [ 285.853s] aws-sdk-ec2

[obi1kenobi]

Thanks for digging into this, I appreciate it! Great work!

I think I'd prefer to leave #767 open though, since this fix addresses the most obvious current problem, but the same issue could just immediately come back on the next lint.

I'll need to find some time to improve the overall query execution around ABIs and the other more complex lints we've started checking since the last performance pass I did. I'll unlink this PR from #767 to remind me.