preserve maxPid in journal aggregation

[alstanchev]

in the rare case where all or most of the elements in an aggregation are pruned for being deleted it is not possible to preserve maxPid which in turn makes it impossible to iterate the whole collection.

the aggregation pipeline:

[{$match: {
 pid: {
  $gt: ''
 }
}}, {$sort: {
 pid: 1,
 sn: -1
}}, {$limit: 200}, {$group: {
 _id: '$pid',
 sn: {
  $first: '$sn'
 },
 __lifecycle: {
  $first: '$s2.__lifecycle'
 },
 _revision: {
  $first: '$s2._revision'
 },
 policyId: {
  $first: '$s2.policyId'
 },
 _modified: {
  $first: '$s2._modified'
 }
}}, {$sort: {
 _id: 1
}}, {$group: {
 _id: null,
 m: {
  $max: '$_id'
 },
 i: {
  $push: '$$ROOT'
 }
}}, {$redact: {
 $cond: {
  'if': {
   $ne: [
    '$__lifecycle',
    'DELETED'
   ]
  },
  then: '$$DESCEND',
  'else': '$$PRUNE'
 }
}}]

Fixes: #1825

[alstanchev]

@thjaeckle would appreciate a review here.
Still waiting on the system test but locally looks good.
On an imported db only one device was added to the search index after the fix all (2k+) were added.

[thjaeckle]

Took a brief look.
Did not deeply look into the how and why this works and the former didn't, but I trust in you and the system tests

[alstanchev]

Tests are green