_node_mass_update_batch_process fails during user cancel when revision is deleted

Created a PR.

Think it needs a rebase showing 1000+ changes

@smustgrave target branch should be 11.0.x or 11.x?

Would let @seanB update the existing branch before we start opening bunch of new ones, since it's only been 1 hour.

Alright @smustgrave

@seanB could you try below code whether it's resolving the problem instead of if condition with instanceOf check.

/** @var \Drupal\node\NodeStorageInterface $storage */
  $storage = \Drupal::entityTypeManager()->getStorage('node');
  $count = min(5, count($context['sandbox']['nodes']));
  for ($i = 1; $i <= $count; $i++) {
    // For each nid, load the node, reset the values, and save it.
    /** @var \Drupal\node\NodeInterface $node */
    $node = array_shift($context['sandbox']['nodes']);
    if ($load) {
      $node = $revisions ?
        $storage->loadRevision($node) : $storage->load($node);
    }
    $node = _node_mass_update_helper($node, $updates, $langcode);

MR!8505 seems quite messed up.

MR!8515 is ready for a review
@seanB could you please check MR!8515 and check whether this resolve the issue.
cc: @smustgrave

Would much rather of given @seanB a chance to update his MR, so now we got 2 one needs to be hidden.

The latest MR does not match the summary.

Tagging for tests based on the title, seems like a bug we should have coverage for.

MR!8515 does not solve the issue. If a node or revision is deleted during the batch process (by something or someone outside of the batch process), $node is NULL. This throws an error when trying to call _node_mass_update_helper.