Proper error handling in JVM bytecode parser

[brianhuffman]

The JVM bytecode parser calls the haskell error function if it ever has any kind of parse error (see e.g. #920). This is not a nice way to handle problems in user-provided input files.

The jvm-parser needs to be redone with proper monadic error handling; reachable code should never call the haskell error function. I notice that it uses the Data.Binary.Get library, which provides two run functions: runGet (which calls error upon failure) and runGetIncremental (which returns a sum type). We should be using runGetIncremental.

[brianhuffman]

Besides the calls to error, the jvm-parser package also does a bunch of array lookups using the partial (!) operator from Data.Array. As the array indices are based on file input, there is no guarantee that they will be in bounds. Therefore we should avoid using this lookup operator.

[brianhuffman]

The uses of error and runGet are addressed by GaloisInc/jvm-parser#9, which is included in saw-script with PR #1074. I think that's enough to close this ticket, although the partial array lookup operator (!) is still in use within jvm-parser, so there's still a bit of room for further improvement.