Persist canonical header references for quick lookup

[i-norden]

canoncial_header() function introduced in #11 can find the canonical header based on the weight in the database, but it is a iterative function that is prohibitively expensive for serving request on-the-fly for headers that are far removed from the current head.

We need to create a trigger function that watches for inserts at the head (n) and uses canoncial_header() to derive and caches the canonical header (or rather a reference to it) for block n - x, where x can be adjusted (start with a default of 2048).

[i-norden]

In the meantime, #25 and #26 have heuristically optimized the canonical_header function

[Burmenska]

@stellarmagnet what about the priority of this task? the last update on it was more than 6 months ago