Update metaopt algorithm

[foxtran]

Originally, during randomisation, fixed atoms could be shifted. Now, they will keep their positions.

Now, normal distribution (with default sigma = 0.1A) is used for sampling structures. In addition, RMSD between structures are also printed.

Original sampling gives the following conformers:

 energy: -930.891821881337 gnorm: 0.000712503100
 energy: -930.891822033756 gnorm: 0.000710760070
 energy: -930.891822128921 gnorm: 0.000708653333
 energy: -930.891822179462 gnorm: 0.000710544110
 energy: -930.891822193101 gnorm: 0.000725795757
 energy: -930.891822262282 gnorm: 0.000719793568
 energy: -930.891822364102 gnorm: 0.000707960824
 energy: -930.891822387881 gnorm: 0.000708387917
 energy: -930.891822380327 gnorm: 0.000719164946
 energy: -930.891822471278 gnorm: 0.000708336494
 energy: -930.891822434372 gnorm: 0.000724310572

With increasing shift to 1e-2, the values are following:

 energy: -930.891821881328 gnorm: 0.000712485608
energy: -930.891820491899 gnorm: 0.000722028810
energy: -930.891821218893 gnorm: 0.000716887212
energy: -930.891823267147 gnorm: 0.000729669729
energy: -930.891822256822 gnorm: 0.000727433638
energy: -930.891818631059 gnorm: 0.000725934166
energy: -930.891816694436 gnorm: 0.000724222055
energy: -930.891814301130 gnorm: 0.000734471975

Switching to normal distribution (sigma = 1e-2):

 energy: -930.891821881329 gnorm: 0.000712485608
energy: -930.891813314260 gnorm: 0.000730103485
energy: -930.891811904553 gnorm: 0.000728859240
energy: -930.891825715596 gnorm: 0.000725408059
energy: -930.891840882407 gnorm: 0.000689468195
energy: -930.891831375352 gnorm: 0.000688700628
energy: -930.891828666761 gnorm: 0.000698186656
energy: -930.891819849089 gnorm: 0.000716279894

And it looks a bit better.

For sigma = 1e-1, it started to work:

 energy: -930.891823290133 gnorm: 0.000708637073
energy: -930.891735088450 gnorm: 0.000792974076
energy: -930.891709831126 gnorm: 0.000831725347
energy: -930.891641709791 gnorm: 0.000815881703
energy: -930.893417885554 gnorm: 0.001999277368
energy: -930.896981612942 gnorm: 0.001618320258
energy: -930.897100611907 gnorm: 0.001457120822

[toxtran]

@awvwgk, i'm going to drop this PR since running of omd took too much time for us. Does anyone have interest on it?

[foxtran]

Oh... Meson builds are broken :|