docs: add comment explaining tolerance for chi logpdf tests

Explains why 40.0 EPS tolerance is needed due to accumulated
floating-point precision differences in complex calculations
involving ln and gammaln functions.

Author: Planeshifter

lib/node_modules/@stdlib/stats/base/dists/chi/logpdf/test/test.native.js

@@ -113,7 +113,7 @@ tape( 'the function evaluates the logpdf for given `x` and `k`', opts, function
 			t.equal( y, expected[ i ], 'x: ' + x[ i ] + ', k: ' + k[ i ] + ', y: ' + y + ', expected: ' + expected[ i ] );
 		} else {
 			delta = abs( y - expected[ i ] );
-			tol = 40.0 * EPS * abs( expected[ i ] );
+			tol = 40.0 * EPS * abs( expected[ i ] ); // Higher tolerance needed due to accumulated floating-point precision differences between JS and C implementations
 			t.ok( delta <= tol, 'within tolerance. x: ' + x[ i ] + '. k: ' + k[ i ] + '. y: ' + y + '. E: ' + expected[ i ] + '. Δ: ' + delta + '. tol: ' + tol + '.' );
 		}
 	}