Improve test coverage

[jcorbin]

Current "100" strategy is misleading, since those numbers were achieve by copious use of ignore exceptions, and never included all package sources.

• make coverage levels honest; this will also make over time analysis, like speculated about in Add coveralls integration #34, more meaningful
• shift to all sources mode, more possible since we've dropped demo code in Drop archery demo build solution #43
• as a temporary measure, disable per-file checking, since only global number are in an enforceable range currently
• drive coverage back up by testing through the command module ala Add a kni command integration test #48
• a kni-test.js mode to cover describe.js Add kni describe mode tests #52
• a kni-test.js mode to cover compiled js mode Cover kni command compiled json mode #53, addressing a latent bug found while linting
• another likely milestone will be to then exclude files like html.js until we can get a headless browser test going
• similarly we'll need to devise an interactive mode for kni-test.js to get coverage of readline.js

[jcorbin]

Here's where we're at under the harsh lens of "no ignores, all sources":

$ npx nyc --all npm test

> [email protected] test
> node -- test.js

-----------------------|---------|----------|---------|---------|-----------------------------------------------------------------------------------------------
File                   | % Stmts | % Branch | % Funcs | % Lines | Uncovered Line #s
-----------------------|---------|----------|---------|---------|-----------------------------------------------------------------------------------------------
All files              |   77.28 |     77.7 |   79.17 |   77.28 |
console.js            |     100 |      100 |     100 |     100 |
document.js           |       0 |        0 |       0 |       0 | 3-154
engine-test.js        |   84.25 |    66.67 |     100 |   84.25 | 113-114,119-120,133-134,137-138,141-142,153-154,167,177-183
engine.js             |   89.43 |    90.57 |    90.2 |   89.61 | 72,76-78,81-82,91,110-111,114-115,286,315-316,371-372,375-376,379-380,427,483,536-547,615-625
evaluate.js           |   98.08 |     96.3 |     100 |   97.98 | 31,39
excerpt.js            |   54.35 |    55.17 |   53.85 |   53.41 | 19,26-46,106-114,148-156,166-200
grammar.js            |   98.11 |    97.44 |   99.29 |   98.11 | 25,43,1112,1137-1138,1158-1159,1293-1294,1316-1317,1387-1388,1601
html.js               |       0 |        0 |       0 |       0 | 3-53
inline-lexer-test.js  |   89.09 |       80 |     100 |   89.09 | 19-23,36
inline-lexer.js       |   99.04 |    98.33 |     100 |   99.04 | 33
kni.js                |       0 |        0 |       0 |       0 | 4-315
link.js               |   94.44 |      100 |     100 |   94.12 | 52-53
outline-lexer-test.js |   84.21 |       75 |     100 |   84.21 | 15-19,32
outline-lexer.js      |   96.97 |    93.75 |     100 |   96.88 | 24
parser.js             |      80 |    66.67 |     100 |      80 | 14,17
path.js               |     100 |      100 |     100 |     100 |
readline.js           |       0 |        0 |       0 |       0 | 3-144
scanner.js            |   96.83 |       90 |     100 |   96.77 | 41,85
scope.js              |     100 |      100 |     100 |     100 |
story.js              |      99 |       50 |     100 |      99 | 16
template.js           |       0 |        0 |       0 |       0 | 3-78
test.js               |     100 |      100 |     100 |     100 |
translate-json.js     |       0 |      100 |       0 |       0 | 2-4
verify.js             |   95.65 |    92.86 |     100 |   95.65 | 57-60
wrapper.js            |      80 |      100 |   71.43 |   82.05 | 16-18,65-68
-----------------------|---------|----------|---------|---------|-----------------------------------------------------------------------------------------------

Without --all we're in the low 90%s ( see #50 )

After #48 we'll at least have coverage of kni.js and maybe others like readline.js

Past that we can add file exclusions for the browser bits, until we can add something like a fake dom or headless browser test.