Retry transient RPC errors on equity price reads

[TaprootFreak]

Problem

TokenService.syncPrices intermittently logs error-level entries when the RPC endpoint has a transient hiccup (observed as recurring bursts on our DEV deployment, ~36/h during an unstable-RPC window):

[TokenService] Failed to update token prices: server response 503 Service Unavailable (... code=SERVER_ERROR, version=6.7.1)
Error: missing revert data (action="call", ...)

Root cause

PriceService.getSpecialTokenPrices reads the equity price directly on the raw provider:

const nativePrice = await equityContract.price();

This is the only on-chain call in the pricing path that bypasses ProviderService's existing transient-error retry policy (withRetry + isTransientRpcError), which getBlock, getBlockNumber, and callBatch already use. A single transient 503 / network blip on this one call fails the whole getTokenPricesInEur cycle.

Fix

• Expose a public ProviderService.call<T>(thunk) that delegates to the existing private withRetry (same policy: 5 attempts, exponential backoff, transient-only — deterministic errors like execution reverted are not retried).
• Route the equity price() read through it.

Retry-then-fail-loud: persistent failures still propagate unchanged to the caller, so this does not mask real outages — fully in line with the fail-loud direction of #70 (which this PR does not overlap; #70 touches the GeckoTerminal/FX paths, this touches getSpecialTokenPrices only).

Verification

• npm run build clean, eslint clean on touched files, prettier clean on price.service.ts (the pre-existing callBatch formatting drift in provider.service.ts is untouched/out of scope).
• Behavior: transient 5xx/timeout/rate-limit → up to 5 retries with backoff; deterministic revert → immediate propagation as before.