Breaking-changes entries owed since 2.5.0: fifteen PRs

[Rafael-SOWNet]

Part of #718.

Six entries that their PRs owed and did not carry:

• Two half-odd powers of secant sums in the half angle's sine #1828 to An odd power of the tangent beside a function of the secant, in the secant #1832: each changed answers, from declined to answered.
• A complex division by a divisor far from unit magnitude is Smith's, which never forms the squared modulus #1374: fixed a wrong one. With the downcasting off, Complex.Create(1.98e119, 0) / Complex.Create(3.39e125, 0) was 0E-100 on 2.5.0.

Every "Was" is measured on a build of 2.5.0 (ab7a8bd6) and every "Now" on master. The answer sizes are the ones the PRs measured. The rows for #1829, #1830's trigonometric one and #1832's second row were probed on 2.5.0 again for this.

The draft notes for 2.6.0 found more candidates: #1368, #1611, #1517, and less certainly #1371, #1377, #1380, #1312, #1289 and #1315. Each needs a probe on 2.5.0 before an entry. They follow separately.

🤖 Generated with Claude Code

https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura

[Rafael-SOWNet]

Added #1368, #1611 and #1517, each reproduced on a build of 2.5.0 (ab7a8bd6) and on master:

• A constant is worth the current precision's digits, and a cached evaluation is recomputed when the precision context has changed #1368: pi at 300 digits after 100 came back with 100 digits on 2.5.0, and 300 now.
• A complex number prints the number it holds, whatever the downcasting setting at printing #1611: 1 - 0.333…i, made with the downcasting off, printed as 1 - 1/3i on 2.5.0.
• The parse cache keeps a parse for each value of the downcasting setting #1517: x^2 + 7/2 reparsed with the downcasting off kept the first parse's Integer and Rational nodes on 2.5.0, and gets Reals now.

#1371, #1377, #1380, #1312, #1289 and #1315 are still to probe.

[Rafael-SOWNet]

Added the other six, each probed on a build of 2.5.0 and on master:

• Euler's substitution, closed: a rational function of x and one square root of a quadratic, rationalised and finished by the rational integrator #1289, The other four trigonometric functions are written as sine and cosine before the complement of a substitution is, and never where that leaves an odd power of the complement under a root #1312, A constant over a linear beside the root of a quadratic is answered by the reciprocal of the linear, and Euler's substitution reaches a radical that is real only between its roots #1315, The inverse trigonometric substitution takes a linear argument, a multiple of its quadratic and a power of a + b arcsin(c x); an exponential of i times an inverse function is algebraic; the resonant exponential times a sine is by Euler; a zero discriminant is seen through its spellings #1380: declined on 2.5.0, answered now. The other four trigonometric functions are written as sine and cosine before the complement of a substitution is, and never where that leaves an odd power of the complement under a root #1312's answer states the condition where it holds.
• A rational function of the hyperbolic tangent with a symbol among its coefficients is integrated by u = tanh(y) with the written factors kept, and an answer assembled from a decomposition with symbols is checked before it is given #1377: sqrt(a + b sech(x)) tanh(x)^5 ran past a minute on 2.5.0, and is answered in 2.4 s now.
• A constant evaluates to a real number whatever the downcasting says, a complex number with a zero imaginary part is real-valued, and the evaluation cache is stamped rather than cleared #1371: with the downcasting off, MathS.e.EvalNumerical() was Complex on 2.5.0, and is Real now.

That clears the list from the notes draft. The 2.6.0 heading rename, from Unreleased, is left for the release itself.