Is your feature request related to a problem? Please describe.

• The IForwardHeaderProvider.GetBlockHeaders returns the currently pending canonical chain that block downloader should download and process.
• This method is called on every peer allocation repeatedly.
• For PowHeaderProvider, which download header directly from best peer, this call is cached, because a p2p call is just straight up too slow.
• For PosHeaderProvider, this was not done for simplicity, preferring to call ChainLevelHelper which re-fetch header.
• This is the reason why there is a hard limit to header lookup in BlockDownloader BlockDownloader.MaxHeaderLookup which limits the parallelism of block downloader. Which, is not nice, but at least much better than downloading from just a single peer.

Describe the solution you'd like

• Cache the result of GetBlockHeaders so that it can be called a few hundreds time per sec without much overhead.
• I don't have a cache invalidation strategy in mind, so that need to be resolved.