perf(legacy-scripting-runner): Optimize LSR imports based on source

[codebycaleb]

Legacy Scripting Runner scripts have some overhead due to the loading of modules that we provide globally (e.g. underscore, moment, etc.). If we optimize the heaviest hitters (moment, jquery, and zfactory), we can see noticeable speed savings (especially for legacy apps that don't use any scripting at all!).

You might notice that this PR doesn't optimize all of the modules we load - just the heaviest hitters. That's because loading some of these takes next to no time at all. Benchmark results from Hyperfine for requiring each one individually (tested against ./nothing.js, a file that consists of module.exports = {};):

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
node test_require.js ./nothing	40.9 ± 1.6	35.8	43.7	1.00
node test_require.js zapier-platform-legacy-scripting-runner/atob	41.9 ± 1.7	37.0	44.3	1.03 ± 0.06
node test_require.js zapier-platform-legacy-scripting-runner/btoa	42.1 ± 1.9	37.6	49.2	1.03 ± 0.06
node test_require.js underscore	42.6 ± 1.9	37.7	46.2	1.04 ± 0.06
node test_require.js crypto	43.4 ± 2.0	40.0	52.7	1.06 ± 0.06
node test_require.js async	43.9 ± 1.8	40.2	47.2	1.08 ± 0.06
node test_require.js xmldom	44.7 ± 1.7	40.9	47.6	1.09 ± 0.06
node test_require.js moment-timezone	51.6 ± 1.6	48.2	55.2	1.26 ± 0.06
node test_require.js zapier-platform-legacy-scripting-runner/zfactory	143.7 ± 3.2	136.9	151.3	3.52 ± 0.16
node test_require.js zapier-platform-legacy-scripting-runner/$	262.5 ± 7.0	255.2	271.7	6.43 ± 0.31