Add possibliity to run a migration by batch

[pp0rtal]

Description

When we do a very long migration we can encounter already:

• slowness: when the migration doesn't rely on an index (COLSCAN, IXSCAN)
• non-verbose: we can't count (noCount:true is set) and we have no relative progress except the bulk log update once ready

In case there are no index, this is more annoying:

• [resumability] a script will always take the same amount of time (scan is what is 99% of the processing)
• [fatal] if there is a socket timeout configured (no match for a long period of time during the scan)

Proposed solution

Batch (avoid timeout, show progression)

We should allow to do the migration progressively relying _id index:

• We iteratively fetch the ranges of _id: .find({_id:{$gte:<previous batch _id limit>}}).sort({_id:1}).skip(<batch size>).limit(1))
• We then apply the migration on the range

Interface: add a new option:

new MongoBulkDataMigration({
    query: // unchanged, might container _id here /!\
    options: {
        throttle: // (already exisst) should throttle over ids
        batchScanSize: 100_000 // default: scan by 100K docs  
    }
})

A batch might result to 0 updates, that's intended.
Ideally, we'd prefer query already set _id to not be overridden, so we should cumulate both .

Resumability