Bound allocation extents for hoist_storage using loop variables one-by-one

[vksnk]

This is a fix for #8141.

As described there, the problem is that bound_of_expr_in_scope is not recursive: when the storage is hoisted, the allocation extents may depend on the loop variables, so we find a maximum bound for them using bound_of_expr_in_scope and the scope which has intervals for each of the loop variables. However, the loop variables themselves may depend on the other loop variables in which case the extent still may refer to loop variables even after calling bound_of_expr_in_scope.

The fix is to apply bound_of_expr_in_scope for each loop variable separately in a reverse order. I wasn't sure if there is a more efficient way to achieve this, but hopefully it isn't so bad, because allocation extents are usually pretty small.

[abadams]

A possibly-easier option might be to do it on the way in, in the for visitor:

    Stmt visit(const For *op) override {
        Expr expanded_min = op->min;
        Expr expanded_extent = op->extent;
        // Iterate from innermost outwards
        for (auto it = hoisted_storages.rbegin(); it != hoisted_storages.rend(); it++) {
            expanded_min = simplify(expand_expr(expanded_min, it->scope));
            expanded_extent = expand_expr(expanded_extent, it->scope);
            Interval loop_bounds = Interval(expanded_min, simplify(expanded_min + expanded_extent - 1));
            loop_bounds.min = bounds_of_expr_in_scope(loop_bounds.min, it->loop_vars).min;
            loop_bounds.max = bounds_of_expr_in_scope(loop_bounds.max, it->loop_vars).max;
            it->loop_vars.push(op->name, loop_bounds);
        }

[vksnk]

A possibly-easier option might be to do it on the way in, in the for visitor:

    Stmt visit(const For *op) override {
        Expr expanded_min = op->min;
        Expr expanded_extent = op->extent;
        // Iterate from innermost outwards
        for (auto it = hoisted_storages.rbegin(); it != hoisted_storages.rend(); it++) {
            expanded_min = simplify(expand_expr(expanded_min, it->scope));
            expanded_extent = expand_expr(expanded_extent, it->scope);
            Interval loop_bounds = Interval(expanded_min, simplify(expanded_min + expanded_extent - 1));
            loop_bounds.min = bounds_of_expr_in_scope(loop_bounds.min, it->loop_vars).min;
            loop_bounds.max = bounds_of_expr_in_scope(loop_bounds.max, it->loop_vars).max;
            it->loop_vars.push(op->name, loop_bounds);
        }

Thanks! They seem to be roughly equivalent w.r.t to number of times bounds_of_expr_in_scope has to be called, so I'll just keep it as is.