change graph on batch verification to comparison of Strauss Piggener

[MaxHillebrand]

This removes the incumbent png file of the improvements of batch verification, which only shows Piggener algorithm [I believe], and adds this graph by @sipa 's research, it shows a comparison of the improvements of of both Strauss and Piggener algorithm, which might be a helpful to display on the rational why Piggener is used in lipsecp256k1.

[sipa]

No, libsecp256k1 implements both Strauss and Pippenger, and picks the optimal one based on size of the batch and available memory. The existing graph shows the actual speedup of batch validation, using actual signature validation code.

The other graph is perhaps interesting too, but I think that's going too low-level into libsecp256k1.

[MaxHillebrand]

Thank you for clearing this up Pieter! In this case, I think that the existing graph is superior, as it shows the actual improvements based on both algorithms.