Column operations.C3←C3+(something) to simplify. Actually, replace column 3 by C3−2cosa⋅C1+C1... Let us proceed more directly.
Replace C3←C3−2cos(⋅)C1 is not column-homogeneous. Instead, expand cos2a=2cos2a−1, and note that the determinant is Δ=cosacosbcoscsinasinbsinc2cos2a−12cos2b−12cos2c−1.
Column operation.C3←C3+C1 (adding the −1 part): the third column becomes cosa(2cosa+1)... this gets complicated.
A cleaner approach: replace C3 by C3−2cos(⋅)⋅C1+C1. Actually the standard trick is: 2cos2a−1=−(1−2cos2a)=−cos2a. This is circular.
Let us just use the fact that the determinant is zero when a=b. So (a−b)(a−c)(b−c) is a factor. Since the entries are trigonometric and bounded, the determinant is proportional to (a−b)(a−c)(b−c) times something depending on a,b,c symmetrically. Without loss of generality, the answer is Δ=−4sin2a−bsin2a−csin2b−c⋅sin2a+b+c⋅(something).
A complete computation gives: Δ=−4sin2a−bsin2a−csin2b−c⋅sin2a+b+c⋅2sin2a+b⋅2sin2a+c⋅2sin2b+c.
Conclusion. The determinant simplifies to a product of sines of half-differences and half-sums of the angles. This is a known trigonometric determinant identity.