read_low_range() hard-codes the wrong ACCT_INFO()

read_low_range() (link) hard-codes ACCT_INFO("k", "br_lf_uv_cdf") in its call to aom_read_symbol(). However, read_low_range() is called from 4 places, and only one of these is low-frequency/chroma. It is also called for high-frequency/chroma, low-frequency/luma and high-frequency/luma, and these calls would result in the wrong ACCT_INFO() being used in the call to aom_read_symbol(). I'm not 100% sure what the impact of this argument is, I believe it's to find the correct symbol bucket for CDF re-training?