Subscript numerals in type error reporting

Motivation

The Futhark compiler reports type errors like so:

Cannot unify `t₅`' with type
  []t₄
as `t₅` must be one of i8, i16, i32, i64, u8, u16, u32, u64, f32, f64 due to integer literal at [6]> :1:19-19).
When matching type
  []t₄
with
  t₅

Proposal

GHC reports type unification using things like a0 or b1. I propose that it instead report not being able to match a₀ with b₁, for example.