Allow M as a variable name

[cscherrer]

Fixes #331

[codecov]

Codecov Report

Merging #340 (5555c69) into master (80dfca9) will increase coverage by 0.05%.
The diff coverage is 41.17%.

@@            Coverage Diff             @@
##           master     #340      +/-   ##
==========================================
+ Coverage   35.10%   35.15%   +0.05%     
==========================================
  Files          62       62              
  Lines        1809     1826      +17     
==========================================
+ Hits          635      642       +7     
- Misses       1174     1184      +10     

Impacted Files	Coverage Δ
src/importance.jl	0.00% <0.00%> (ø)
src/particles.jl	0.00% <0.00%> (ø)
src/primitives/entropy.jl	0.00% <0.00%> (ø)
src/primitives/insupport.jl	0.00% <0.00%> (ø)
src/primitives/likelihood-weighting.jl	0.00% <0.00%> (ø)
src/primitives/simulate.jl	0.00% <0.00%> (ø)
src/symbolic/symbolic.jl	0.00% <0.00%> (ø)
src/primitives/as.jl	61.70% <100.00%> (+0.83%)	⬆️
src/primitives/basemeasure.jl	89.47% <100.00%> (+0.58%)	⬆️
src/primitives/logdensity.jl	88.88% <100.00%> (+0.65%)	⬆️
... and 2 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 80dfca9...5555c69. Read the comment docs.

[cscherrer]

The original code has let M without a binding. I'm still not sure of the semantics of that. This PR works for the example, but I'm not sure it's actually correct, for example it may have edge cases when working across modules.