Add prototype Github CI workflow

p^(2d-2) algorithm for L2L

See merge request inducer/sumpy!123

Asymptotically better algorithm for M2M (In both compressed and full)

See merge request !122

Some readability improvements for the new M2M algorithm

See merge request isuruf/sumpy!2

On master, M2M is $`O(p^2d)`$ for order p and dimension d, but CSE
reduces this down to $`O(p^{2d-1})`$ sometimes.
For eg: in Helmholtz 2D, full taylor is $`O(p^{2d-1})`$ and
HelmholtzConformingTaylor is $`O(p^{2d-1.5})`$.

This commit produces expressions in $`O(p^{2d-1})`$ consistently
regardless of how good CSE is and $`O(p^{2d-2})`$ for compressed.

The new algorithm uses the observation that M2M coefficients
have the form in 2D,

$`B_{m, n} = \sum_{i\le m, j\le n} A_{i, j} d_x^i d_y^j \binom{m}{i} \binom{n}{j}`$

and can be rewritten as follows,

Let $`T_{m, n} = \sum_{i\le m} A_{i, n} d_x^i \binom{m}{i}`$.

Then, $`B_{m, n} = \sum_{j\le n} T_{m, j} d_y^j \binom{n}{j}`$
and $`T_{m, n}`$ are $`p^2`$ number of temporary variables that are
reused for different M2M coefficients and costs $`p`$ per variable.
Total cost for calculating $`T_{m, n}`$ is $`p^3`$ and similar for $`B_{m, n}`$

Add Biharmonic 3D

See merge request !121

Biharmonic kernel fixes

See merge request !120

This reverts commit 7cf3197b.

Fix get_source_args for DirectionalTargetDerivative

See merge request inducer/sumpy!119