improve stacktrace when dangerfile failed to run

[sibelius]

for this dangerfile.js

import { danger, warn, message } from 'danger';
import fs from 'fs';

// TODO - improve dangerjs

const modifiedMD = danger.git.modified_files.join('\n- ');

const allFiles = danger.git.modified_files.concat(danger.git.created_files);

const includesEdge = allFiles.map(file => {
  const reAdd = /\BAdd|Add\B/g;
  const reEdge = /\BEdge|Edge\B/g;
  const content = fs.readFileSync(file).toString();
  const fileHasAMutationAdd = file.match(reAdd);
  const fileHaveAnEdge = content.match(reEdge);
  fileHasAMutationAdd &&
    !fileHaveAnEdge &&
    warn(`This file ${file} seems to be an add mutation and not include an output Edge`);
});

message(`Changed Files in this PR: \n - ${modifiedMD}`);

it is throwing the following error stacktrace:

ENOENT: no such file or directory, open '.github/PULL_REQUEST_TEMPLATE.md'
Error: ENOENT: no such file or directory, open '.github/PULL_REQUEST_TEMPLATE.md'
    at Object.fs.openSync (fs.js:660:18)
    at Object.fs.readFileSync (fs.js:565:33)
    at Object.apply (/app/node_modules/vm2/lib/contextify.js:288:34)
    at /app/danger-0.jelc7b1175j.js:18:30
    at Object.apply (/app/node_modules/vm2/lib/contextify.js:87:36)
    at Array.map (<anonymous>)
    at Object.apply (/app/node_modules/vm2/lib/contextify.js:288:34)
    at Object.<anonymous> (/app/danger-0.jelc7b1175j.js:14:29)
    at NodeVM.run (/app/node_modules/vm2/lib/main.js:427:27)
    at Object.<anonymous> (/app/node_modules/danger/distribution/runner/runners/vm2.js:97:37)
    at step (/app/node_modules/danger/distribution/runner/runners/vm2.js:40:23)
    at Object.next (/app/node_modules/danger/distribution/runner/runners/vm2.js:21:53)
    at /app/node_modules/danger/distribution/runner/runners/vm2.js:15:71
    at new Promise (<anonymous>)
    at __awaiter (/app/node_modules/danger/distribution/runner/runners/vm2.js:11:12)
    at Object.exports.runDangerfileEnvironment (/app/node_modules/danger/distribution/runner/runners/vm2.js:80:121)

it'd be great to show where in dangerfile the error was thrown

[orta]

This problem only happens on self hosted instances using vm2, and shouldn't be an issue in peril staging (where evaluation is a 90% closer to normally running a node script) so I'm unlikely to take a stab at this

[sibelius]

this happened on peril (on heroku) running a dangerfile