Improve test harness diagnostics

[takikawa]

This PR improves debugging with the spec tests. For invalid source maps it prints the failure reason so it's much easier to check if maps are rejected for the right reason.

For known failures, use todo to make it easier to track why it's failing.

Also update the spec test submod.

Note: in most cases, invalid maps are being rejected for the right reason thankfully. But in some cases the test cases need tweaking to avoid the wrong exception (there's a fairly trivial issue related to source handling) and there are a few cases where the upstream source-map library is not error-checking where it should. I'll make separate PRs related to these.

[takikawa]

To illustrate this PR a bit better, here's a sample of how it shows invalid SM tests:

  ✔ The source map spec test case "validMappingEmptyGroups" has a valid source map (0.882309ms)                                                                           
  ✔ The source map spec test case "validMappingEmptyString" has a valid source map (0.980224ms)                                                                           
  ✔ The source map spec test case "indexMapWrongTypeSections" has an invalid source map (0.765349ms)                                                                      
  ℹ Source map "sections" field is not an object.                                                                                                                         
  ✔ The source map spec test case "indexMapWrongTypeOffset" has an invalid source map (0.896847ms)                                                                        
  ℹ Source file not found: empty-original.js undefined                                                                                                                    
  ✔ The source map spec test case "indexMapWrongTypeMap" has an invalid source map (0.818119ms)                                                           
  ℹ Index map section 0 is missing a "map" field or it is invalid.   

Before this PR the ℹ lines weren't present so you couldn't tell why it was rejected. In the second to last case, the exception is actually spurious which I'll fix later.

[jkup]

This is great, thanks!