A power of a product with a sum in it is multiplied out as a polynomial

[Rafael-SOWNet]

Part of #718.

A power of a product with a sum in it is multiplied out as a polynomial. In ExpandedOverTheSymbols, the fast path, which multiplies out over the symbols as a polynomial, was taken only for a sum raised to a power or a product with sums on both sides. Beside (d + k x)^3, the norm of the other factors' residue modulo a repeated quadratic is (S c)^2, where S has 28 terms. That is neither shape, so it went to Entity.Expand. That one expansion took 27 of the 29 seconds of 1/((d + k x)^3 (a + b x + c x^2)^2). The check now accepts a whole power of anything that contains a sum.

	2.5.0	master 52ce4d8	branch
1/((d + k x)^3 (a + b x + c x^2)^2)	declined	9,593 chars, 29.8 s	the same answer, byte for byte, 1.4 s
1/((d + k x)(f + g x)(a + b x + c x^2)^2)	declined	no answer within 300 s	24,143 chars, 1.6 s

Measured against master 52ce4d8

• Rubi's 1.2 files, 3,038 rows: 3,025 solved on master against 3,027 on the branch, with 0 wrong. The two gains are 1.2.1.2:2493 and 1.2.1.4:1051, both timeouts on master.
• The full chain over families 0–8: 3 rows moved. Two are those gains. The third, 1.1.3.4:857, runs about 25 s on both builds, which is on the budget's boundary. 0 rows are wrong.
• Gate: passed.
• Suite: 15,244 passed, 0 failed.

SymbolicFactorsPartialFractionsTest gains a row that differentiates the first integral back and bounds its size.

The other timeouts in 1.2.1.2, with the quadratic to the fourth and fifth power, are a different cost and are not addressed here.

🤖 Generated with Claude Code

https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura