dgb(tracker): derive chain-score time_span from CoinParams::block_period (drop farsider350 holdover)

[frstrtr]

Stacked on #131 (dgb/oracle-conformance).

What

best-chain score() computed time_span = block_height * 15 with a hardcoded 15s period and a dead farsider350 digibyte.py:21 comment. That bypassed the oracle-conformed PARENT.BLOCK_PERIOD (75s, config_coin.hpp set in #131). This sources it from the make_coin_params-populated m_params->block_period so there is ONE SSOT for the DGB Scrypt-only parent period in the score path — no hardcoded consensus holdover. Stale confirmation comments updated (1=tip -> 75s, ~75M fallback).

Bucket (v36 standardization)

PRE-V36 holdover removal -> conform to own oracle frstrtr/p2pool-dgb-scrypt. Coin-specific consensus value; per-coin, not a cross-coin standardization target.

Scope / safety

• dgb-only (src/impl/dgb/share_tracker.hpp); no shared-base / bitcoin_family touch.
• c2pool-dgb builds clean (EXIT=0, binary links). NOTE: share_tracker.hpp is header-only and not yet pulled into a compiled TU (node is still the skeleton; the pool layer that instantiates the tracker lands in the stacked pool/share slices) — so the line is not compile-exercised by the current smoke. It is valid in the context it will compile (m_params is an established member, block_period a CoinParams field).

NO AI/model attribution. GPG-signed (c0ea7c7).

[frstrtr]

Superseded by #145 (DGB Phase-B batch land #132->#142, operator-approved merge-as-reviewed; content incl. this PR verified present in the batch). Landed to master via #145.