another cases of Bach constant cheating

[Igor Schein <igor@txc.com>]

Hi,

? gettime;forstep(k=2.,.3,-.1,print1(k" ");setrand(1579152476);bnr=bnrinit(bnfinit(quadpoly(13829,y),,[k,k]*1.),211,1);rnfkummer(bnr,[5,1;0,1]);print(round(gettime/10^3)"s"))
2.000000000000000000000000000 34s
1.900000000000000000000000000 34s
1.800000000000000000000000000 34s
1.700000000000000000000000000 20s
1.600000000000000000000000000 20s
1.500000000000000000000000000 20s
1.400000000000000000000000000 20s
1.300000000000000000000000000 20s
1.200000000000000000000000000 20s
1.100000000000000000000000000 20s
1.000000000000000000000000000 20s
0.9000000000000000000000000002 36s
0.8000000000000000000000000002
  ***   user interrupt after 33mn, 28,380 ms.

If bound is less then 0.9, rnfkummer() enters an infinite loop ( true
infinite loop, no doubling the bound ).  Also, timing pattern is a bit
peculiar, 36s at 0.9.

Thanks

Igor