lint/mypy: silence more messages on "improper" f-string formatting.

Copy-paste from the earlier commit (that was cherry-picked from
previous PR):

The messages silenced in this commit are all:

    On Python 3 '{}'.format(b'abc') produces "b'abc'"; use !r if this is a desired behavior

Sadly, `mypy` doesn't say which particular fields are problematic; I've
tried my best to single them out - hopefully, nothing extraneous got in.

Author: veox

eth/db/chain.py

@@ -141,16 +141,16 @@ def persist_unexecuted_block(self,
 
         if tx_root_hash != block.header.transaction_root:
             raise ValidationError(
-                f"Block's transaction_root ({block.header.transaction_root}) "
-                f"does not match expected value: {tx_root_hash}"
+                f"Block's transaction_root ({block.header.transaction_root!r}) "
+                f"does not match expected value: {tx_root_hash!r}"
             )
 
         receipt_root_hash, receipt_kv_nodes = make_trie_root_and_nodes(receipts)
 
         if receipt_root_hash != block.header.receipt_root:
             raise ValidationError(
-                f"Block's receipt_root ({block.header.receipt_root}) "
-                f"does not match expected value: {receipt_root_hash}"
+                f"Block's receipt_root ({block.header.receipt_root!r}) "
+                f"does not match expected value: {receipt_root_hash!r}"
             )
 
         with self.db.atomic_batch() as db:

eth/db/header.py

@@ -261,7 +261,7 @@ def _persist_checkpoint_header(
                     # True parent should have already been canonicalized during
                     #   _set_as_canonical_chain_head()
                     raise ValidationError(
-                        f"Why was a non-matching parent header {parent_hash} left as canonical "
+                        f"Why was a non-matching parent header {parent_hash!r} left as canonical "
                         f"after _set_as_canonical_chain_head() and {true_parent} is available?"
                     )