Better support for --sync.loop.block.limit

[AskAlexSharov]

Now if set --sync.loop.block.limit=100:

• node will download all blocks to db (likely to tmpdb and then copy to db)
• don't mark blocks as canonical
• and then will start process them by 100 - only because StageSenders has support for this flag

Need better support for --sync.loop.block.limit because E3 sync speed depends on DB size. And to keep it small - better don't download all blocks.

Read also: #10286 (comment)

• Do we need support by Caplin? Or it must be only EL feature?
• To re-visit and simplify updateForkChoice function. In E2 we did revert support of this flag in updateForkChoice. But in E3 we still have (because hard to revert).

[Giulio2002]

Took care of it in: #10341