tgsi: fix incorrect tgsi_shader_info::num_tokens computation
We were incrementing num_tokens in each loop iteration while parsing
the shader. But each call to tgsi_parse_token() can consume more than
one token (and often does). Instead, just call the tgsi_num_tokens()
Luckily, this issue doesn't seem to effect any current users of this
field (llvmpipe just checks for <= 1, for example).
Reviewed-by: Neha Bhende<email@example.com>
Reviewed-by: Roland Scheidegger <firstname.lastname@example.org>