Fix test cases: if hf 343 and 890 occur at same maintenance interval

[abitmore]

The test case hf_935_test was failing in testnet-hf-time branch when bitasset data is updated before hard fork 890:

bitshares-core/tests/tests/bitasset_tests.cpp

Line 1071 in e9defea

BOOST_CHECK( db.find<limit_order_object>( sell_id ) );

The reason is explained in the in-code comments:

bitshares-core/libraries/chain/db_maint.cpp

Lines 933 to 935 in e9defea

	// Note: due to bitshares-core issue #935, the check below (using median_changed) is incorrect.
	// However, `skip_check_call_orders` will likely be true in both testnet and mainnet,
	// so effectively the incorrect code won't make a difference.

That said, if hf 343 and 890 occur at same maintenance interval, after median feed got updated by hf 890 code, check_call_orders will be called by hf 343 code, thus will finally produce a good result.

This PR fixed the test case.

[abitmore]

Pushed a new commit that fixes an unsigned integer underflow issue.