Assuming that there is a missing blank line between your first two records, then a one-liner with 'paragraph mode' will do the trick:

perl -e"BEGIN{$/=''}" -ple"tr[\n][]d" input.file > output.file
[download]

It looks like your data is delimited by double returns. However, your example breaks that pattern for the first record.

Therefore the below solution assumes that paragraph mode wouldn't work:

my $data = do {local $/; <DATA>};

$data =~ s/\n(?=\=\?|\+)//g;

print $data;

__DATA__
Xmodels node1 node2 D G S S Al=1 X=2 V=2e-6
=? bitval=1 model=bitval20 color=30 
=? coordinates x-val=20 y-val=30 targetDestroyed=yes
=? missionstatus=completed perimeterSecured=yes
=? repeatMission=false newMission=yes
=? ordinance=34!44 countrycode=red
Xmodels node1020 node4300 D G S S Al=12 X=222 V=23e-6
=? bitval=0101 model=bitval28 color=101AFFFB 
=? coordinates x-val=2370 y-val=390 targetDestroyed=no
=? missionstatus=abandoned perimeterSecured=indeterminate
=? repeatMission=false newMission=InexploreStage
=? ordinance=NULL countrycode=purple

Mmodel 300 90 axis-x axis-y 10110001
+ firemethod=automatic burstmode=yes
+ modeltype=gatlin revolve=true
+ mmodellinespeed=3000fps targacquire=manual manned=optional

Mmodel 330 190 axis-x axis-y 1A11FFF1
+ firemethod=automatic burstmode=yes
+ modeltype=FOEAWE revolve=true
+ mmodellinespeed=3000fps targacquire=automatic
+ manned=optional
+ roundspersecond=13000/s
[download]

use warnings;
use strict;
while (<DATA>) {
    s/\n/ /;
    s/^(\wmodel)/\n\n$1/;
    print;
}

__DATA__
Xmodels node1 node2 D G S S Al=1 X=2 V=2e-6
=? bitval=1 model=bitval20 color=30 
=? coordinates x-val=20 y-val=30 targetDestroyed=yes
=? missionstatus=completed perimeterSecured=yes
=? repeatMission=false newMission=yes
=? ordinance=34!44 countrycode=red
Xmodels node1020 node4300 D G S S Al=12 X=222 V=23e-6
=? bitval=0101 model=bitval28 color=101AFFFB 
=? coordinates x-val=2370 y-val=390 targetDestroyed=no
=? missionstatus=abandoned perimeterSecured=indeterminate
=? repeatMission=false newMission=InexploreStage
=? ordinance=NULL countrycode=purple

Mmodel 300 90 axis-x axis-y 10110001
+ firemethod=automatic burstmode=yes
+ modeltype=gatlin revolve=true
+ mmodellinespeed=3000fps targacquire=manual manned=optional

Mmodel 330 190 axis-x axis-y 1A11FFF1
+ firemethod=automatic burstmode=yes
+ modeltype=FOEAWE revolve=true
+ mmodellinespeed=3000fps targacquire=automatic
+ manned=optional
+ roundspersecond=13000/s
[download]

When I consider problems like the one you describe, I freely admit that I’m heavily influenced by a tool ... awk ... that is known to be one of the (so to speak...) “inspirations for” Perl.   This tool (IMHO) encourages you to consider a file “line-type by line-type.”   It is a very-useful, generalized approach to problems of this type that generally consists of three questions:

1. What type of line (record...) am I right-now looking at?
2. Now that I know this ... what should I do with this type of line?
3. Now that I’ve done this ... “am I finished yet?”   (And if so, what should I do?)

Bear with me, now...   You might, for example, decide that the “types of lines” might be:

• (First line... beginning of file...)
• Lines that begin with XModel
• Lines that begin with model
• Lines that begin with =?\b
• Blank lines
• (Last line... end of file...)

Please let me refer you to man awk from this point forward.   awk is, in fact, a generalized tool to address this very type of problem, and it is one of the acknowledged inspirations for Perl.   No matter what tool you use to solve your problem, the approach that is encouraged by awk is (I think...) very useful food for thought.