Handle x86 errors more gracefully

[atomb]

And provide a template for catching other varieties of exception more
consistently.

Fixes #678.

[RyanGlScott]

Is the idea to consolidate all of the exception-handling code into the io function? I ask since there are other places where liftIO is used in conjunction with catch:

saw-script/src/SAWScript/Crucible/JVM/Override.hs

Line 971 in 46249db

liftIO $ resolveSetupVal cc m tyenv nameEnv sval' `X.catch` handleException opts

saw-script/src/SAWScript/Crucible/LLVM/Override.hs

Line 1845 in aac0a02

lval <- liftIO $ resolveSetupVal cc mem m tyenv nameEnv sval' `X.catch` handleException opts

Would it make sense to consolidate these bits of functionality into io as well?

[atomb]

I think it would make sense, unless anyone else knows of issues that might come up as a result of that. And there are almost 300 instances of liftIO in the saw-script repo that could probably be migrated. My current thought is that this PR lays the groundwork for doing that, but that we make the more sweeping changes after the 0.8 release.

[RyanGlScott]

That sounds reasonable to me. I suppose the only other thing I might desire from this PR is a test case that demonstrates the new error message, assuming that that is feasible to set up. If that's not feasible, then perhaps showing an example of the new error message in a GitHub comment would suffice.

[atomb]

Our current test infrastructure would make it a bit difficult to check the error message in this case. It would be nice to set something up to make that straightforward in the future, though. For the moment, here's an example of the error message I get from the code in this PR when I change the ELF text in an executable header to EGF:

[19:11:50.679] Loading file "/Users/atomb/galois/saw-script/intTests/test_llvm_x86_01/test.saw"
[19:11:50.684] Finding symbol for "foo"
[19:11:50.684] Stack trace:
"llvm_verify_x86" (/Users/atomb/galois/saw-script/intTests/test_llvm_x86_01/test.saw:14:20-14:35):
Error in x86 code: Invalid ELF header at offset 0: Invalid magic number for ELF: ("\DELEGF","\DELELF")

[RyanGlScott]

Excellent. We should open a separate issue about consolidating exception-handler–related code in io, but beyond that, LGTM.