add timeout to postgres backup commands

Kill pg_dump/pg_restore after 30 minutes so a lock wait cannot hang a
backup or leave the server disconnected from its database during a
restore. A killed restore rolls back via --single-transaction.

server/managers/BackupManager.js

@@ -657,6 +657,8 @@ class BackupManager {
 
       const options = {
         maxBuffer: 10 * 1024 * 1024,
+        // Kill after 30 mins (e.g. lock waits) - a killed restore rolls back via --single-transaction
+        timeout: 30 * 60 * 1000,
         env: connection.env
       }
       childProcess.execFile(command, [...args, ...connection.args], options, (error, stdout, stderr) => {

test/server/managers/BackupManager.test.js

@@ -76,6 +76,7 @@ describe('BackupManager', () => {
     ])
     expect(execFileStub.firstCall.args[1].join(' ')).to.not.include('secretpass')
     expect(execFileStub.firstCall.args[2].env.PGPASSWORD).to.equal('secretpass')
+    expect(execFileStub.firstCall.args[2].timeout).to.be.a('number')
   })
 
   it('should restore Postgres dumps in one transaction and clean existing objects', async () => {