Mongo Add account name indexes to query data faster

[maslenitsa93]

Analysts are often using Mongo aggregation queries with $lookup, like this one:

db.account_object.aggregate([{
$match: {
}}, {
    $lookup: {
        "from": "comment_object",
        let: {
            authorr: "$name"
        },
        pipeline: [{$match: {$expr: {$and: [{"$eq": ["$author", "$$authorr"]}]}}}],
        "as": "his_comments"
    }
}])

Such queries are very slow. Let's run such query when we have ~590.000 comments. Result: 15:32 - Started 15:38 - Ended

It is so slow because we using comment_object.author field which is just string, and not indexed ObjectId field.

Let's add comment_object.author_id field with index on it, and run slightly rewrited query after full re-sync:

db.account_object.aggregate([{
$match: {
}}, {
    $lookup: {
        "from": "comment_object",
        let: {
            authorr: "$_id"
        },
        pipeline: [{$match: {$expr: {$and: [{"$eq": ["$author_id", "$$authorr"]}]}}}],
        "as": "his_comments"
    }
}])

Result: 15:43:02 - Started 15:43:03 - Ended

Indexes could strongly improve performance of using & development of Mongo aggregation queries.

In this issue:

• Let's add them to all account field in all often-using collections.
• (Optionally) Improve code of adding of index. It is too long - 3 lines:

          document owner_index;
          owner_index << "owner_id" << 1;
          doc.indexes_to_create.push_back(std::move(owner_index));

  It is done for more flexibility, but usually we don't need to index on multiple fields. Let's write helper function which will allow to create such index in 1 line of code (with 2 args - field name and index value, i.e. "owner_id" and 1), but if need we'll still could do it with an old way.

[afalaleev]

Looks good. But decision is not required, now.