Docs say sum() and mean() are full-reduction only

[Terminay]

docs/api-reference/tensor.md lines 77 and 80 describe sum() and mean() as "full reduction, not per-axis".

This is incorrect, both methods accept axis and keepdims parameters and do support per-axis reduction.

[github-actions]

The issue points out that the documentation for sum() and mean() in docs/api-reference/tensor.md inaccurately describes these methods as "full reduction, not per-axis," while in reality, both methods support per-axis reduction through the axis and keepdims parameters.